diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000000000000000000000000000000000000..ff1446e90bffc5485e346d9fd8eefdeb28cdf646
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,146 @@
+.idea/
+
+training/
+lightning_logs/
+image_log/
+
+*.pth
+*.pt
+*.ckpt
+*.safetensors
+
+gradio_pose2image_private.py
+gradio_canny2image_private.py
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# dlib
+**/*.dat
diff --git a/annotator/canny/__init__.py b/annotator/canny/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..cb0da951dc838ec9dec2131007e036113281800b
--- /dev/null
+++ b/annotator/canny/__init__.py
@@ -0,0 +1,6 @@
+import cv2
+
+
+class CannyDetector:
+    def __call__(self, img, low_threshold, high_threshold):
+        return cv2.Canny(img, low_threshold, high_threshold)
diff --git a/annotator/ckpts/ckpts.txt b/annotator/ckpts/ckpts.txt
new file mode 100644
index 0000000000000000000000000000000000000000..1978551fb2a9226814eaf58459f414fcfac4e69b
--- /dev/null
+++ b/annotator/ckpts/ckpts.txt
@@ -0,0 +1 @@
+Weights here.
\ No newline at end of file
diff --git a/annotator/hed/__init__.py b/annotator/hed/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..56532c374df5c26f9ec53e2ac0dd924f4534bbdd
--- /dev/null
+++ b/annotator/hed/__init__.py
@@ -0,0 +1,132 @@
+import numpy as np
+import cv2
+import os
+import torch
+from einops import rearrange
+from annotator.util import annotator_ckpts_path
+
+
+class Network(torch.nn.Module):
+    def __init__(self, model_path):
+        super().__init__()
+
+        self.netVggOne = torch.nn.Sequential(
+            torch.nn.Conv2d(in_channels=3, out_channels=64, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=64, out_channels=64, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False)
+        )
+
+        self.netVggTwo = torch.nn.Sequential(
+            torch.nn.MaxPool2d(kernel_size=2, stride=2),
+            torch.nn.Conv2d(in_channels=64, out_channels=128, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=128, out_channels=128, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False)
+        )
+
+        self.netVggThr = torch.nn.Sequential(
+            torch.nn.MaxPool2d(kernel_size=2, stride=2),
+            torch.nn.Conv2d(in_channels=128, out_channels=256, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=256, out_channels=256, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=256, out_channels=256, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False)
+        )
+
+        self.netVggFou = torch.nn.Sequential(
+            torch.nn.MaxPool2d(kernel_size=2, stride=2),
+            torch.nn.Conv2d(in_channels=256, out_channels=512, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False)
+        )
+
+        self.netVggFiv = torch.nn.Sequential(
+            torch.nn.MaxPool2d(kernel_size=2, stride=2),
+            torch.nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False),
+            torch.nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),
+            torch.nn.ReLU(inplace=False)
+        )
+
+        self.netScoreOne = torch.nn.Conv2d(in_channels=64, out_channels=1, kernel_size=1, stride=1, padding=0)
+        self.netScoreTwo = torch.nn.Conv2d(in_channels=128, out_channels=1, kernel_size=1, stride=1, padding=0)
+        self.netScoreThr = torch.nn.Conv2d(in_channels=256, out_channels=1, kernel_size=1, stride=1, padding=0)
+        self.netScoreFou = torch.nn.Conv2d(in_channels=512, out_channels=1, kernel_size=1, stride=1, padding=0)
+        self.netScoreFiv = torch.nn.Conv2d(in_channels=512, out_channels=1, kernel_size=1, stride=1, padding=0)
+
+        self.netCombine = torch.nn.Sequential(
+            torch.nn.Conv2d(in_channels=5, out_channels=1, kernel_size=1, stride=1, padding=0),
+            torch.nn.Sigmoid()
+        )
+
+        self.load_state_dict({strKey.replace('module', 'net'): tenWeight for strKey, tenWeight in torch.load(model_path).items()})
+
+    def forward(self, tenInput):
+        tenInput = tenInput * 255.0
+        tenInput = tenInput - torch.tensor(data=[104.00698793, 116.66876762, 122.67891434], dtype=tenInput.dtype, device=tenInput.device).view(1, 3, 1, 1)
+
+        tenVggOne = self.netVggOne(tenInput)
+        tenVggTwo = self.netVggTwo(tenVggOne)
+        tenVggThr = self.netVggThr(tenVggTwo)
+        tenVggFou = self.netVggFou(tenVggThr)
+        tenVggFiv = self.netVggFiv(tenVggFou)
+
+        tenScoreOne = self.netScoreOne(tenVggOne)
+        tenScoreTwo = self.netScoreTwo(tenVggTwo)
+        tenScoreThr = self.netScoreThr(tenVggThr)
+        tenScoreFou = self.netScoreFou(tenVggFou)
+        tenScoreFiv = self.netScoreFiv(tenVggFiv)
+
+        tenScoreOne = torch.nn.functional.interpolate(input=tenScoreOne, size=(tenInput.shape[2], tenInput.shape[3]), mode='bilinear', align_corners=False)
+        tenScoreTwo = torch.nn.functional.interpolate(input=tenScoreTwo, size=(tenInput.shape[2], tenInput.shape[3]), mode='bilinear', align_corners=False)
+        tenScoreThr = torch.nn.functional.interpolate(input=tenScoreThr, size=(tenInput.shape[2], tenInput.shape[3]), mode='bilinear', align_corners=False)
+        tenScoreFou = torch.nn.functional.interpolate(input=tenScoreFou, size=(tenInput.shape[2], tenInput.shape[3]), mode='bilinear', align_corners=False)
+        tenScoreFiv = torch.nn.functional.interpolate(input=tenScoreFiv, size=(tenInput.shape[2], tenInput.shape[3]), mode='bilinear', align_corners=False)
+
+        return self.netCombine(torch.cat([ tenScoreOne, tenScoreTwo, tenScoreThr, tenScoreFou, tenScoreFiv ], 1))
+
+
+class HEDdetector:
+    def __init__(self):
+        remote_model_path = "https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/network-bsds500.pth"
+        modelpath = os.path.join(annotator_ckpts_path, "network-bsds500.pth")
+        if not os.path.exists(modelpath):
+            from basicsr.utils.download_util import load_file_from_url
+            load_file_from_url(remote_model_path, model_dir=annotator_ckpts_path)
+        self.netNetwork = Network(modelpath).cuda().eval()
+
+    def __call__(self, input_image):
+        assert input_image.ndim == 3
+        input_image = input_image[:, :, ::-1].copy()
+        with torch.no_grad():
+            image_hed = torch.from_numpy(input_image).float().cuda()
+            image_hed = image_hed / 255.0
+            image_hed = rearrange(image_hed, 'h w c -> 1 c h w')
+            edge = self.netNetwork(image_hed)[0]
+            edge = (edge.cpu().numpy() * 255.0).clip(0, 255).astype(np.uint8)
+            return edge[0]
+
+
+def nms(x, t, s):
+    x = cv2.GaussianBlur(x.astype(np.float32), (0, 0), s)
+
+    f1 = np.array([[0, 0, 0], [1, 1, 1], [0, 0, 0]], dtype=np.uint8)
+    f2 = np.array([[0, 1, 0], [0, 1, 0], [0, 1, 0]], dtype=np.uint8)
+    f3 = np.array([[1, 0, 0], [0, 1, 0], [0, 0, 1]], dtype=np.uint8)
+    f4 = np.array([[0, 0, 1], [0, 1, 0], [1, 0, 0]], dtype=np.uint8)
+
+    y = np.zeros_like(x)
+
+    for f in [f1, f2, f3, f4]:
+        np.putmask(y, cv2.dilate(x, kernel=f) == x, x)
+
+    z = np.zeros_like(y, dtype=np.uint8)
+    z[y > t] = 255
+    return z
diff --git a/annotator/midas/__init__.py b/annotator/midas/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..dc5ac03eea6f5ba7968706f1863c8bc4f8aaaf6a
--- /dev/null
+++ b/annotator/midas/__init__.py
@@ -0,0 +1,38 @@
+import cv2
+import numpy as np
+import torch
+
+from einops import rearrange
+from .api import MiDaSInference
+
+
+class MidasDetector:
+    def __init__(self):
+        self.model = MiDaSInference(model_type="dpt_hybrid").cuda()
+
+    def __call__(self, input_image, a=np.pi * 2.0, bg_th=0.1):
+        assert input_image.ndim == 3
+        image_depth = input_image
+        with torch.no_grad():
+            image_depth = torch.from_numpy(image_depth).float().cuda()
+            image_depth = image_depth / 127.5 - 1.0
+            image_depth = rearrange(image_depth, 'h w c -> 1 c h w')
+            depth = self.model(image_depth)[0]
+
+            depth_pt = depth.clone()
+            depth_pt -= torch.min(depth_pt)
+            depth_pt /= torch.max(depth_pt)
+            depth_pt = depth_pt.cpu().numpy()
+            depth_image = (depth_pt * 255.0).clip(0, 255).astype(np.uint8)
+
+            depth_np = depth.cpu().numpy()
+            x = cv2.Sobel(depth_np, cv2.CV_32F, 1, 0, ksize=3)
+            y = cv2.Sobel(depth_np, cv2.CV_32F, 0, 1, ksize=3)
+            z = np.ones_like(x) * a
+            x[depth_pt < bg_th] = 0
+            y[depth_pt < bg_th] = 0
+            normal = np.stack([x, y, z], axis=2)
+            normal /= np.sum(normal ** 2.0, axis=2, keepdims=True) ** 0.5
+            normal_image = (normal * 127.5 + 127.5).clip(0, 255).astype(np.uint8)
+
+            return depth_image, normal_image
diff --git a/annotator/midas/api.py b/annotator/midas/api.py
new file mode 100644
index 0000000000000000000000000000000000000000..1ab9f15bf96bbaffcee0e3e29fc9d3979d6c32e8
--- /dev/null
+++ b/annotator/midas/api.py
@@ -0,0 +1,169 @@
+# based on https://github.com/isl-org/MiDaS
+
+import cv2
+import os
+import torch
+import torch.nn as nn
+from torchvision.transforms import Compose
+
+from .midas.dpt_depth import DPTDepthModel
+from .midas.midas_net import MidasNet
+from .midas.midas_net_custom import MidasNet_small
+from .midas.transforms import Resize, NormalizeImage, PrepareForNet
+from annotator.util import annotator_ckpts_path
+
+
+ISL_PATHS = {
+    "dpt_large": os.path.join(annotator_ckpts_path, "dpt_large-midas-2f21e586.pt"),
+    "dpt_hybrid": os.path.join(annotator_ckpts_path, "dpt_hybrid-midas-501f0c75.pt"),
+    "midas_v21": "",
+    "midas_v21_small": "",
+}
+
+remote_model_path = "https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/dpt_hybrid-midas-501f0c75.pt"
+
+
+def disabled_train(self, mode=True):
+    """Overwrite model.train with this function to make sure train/eval mode
+    does not change anymore."""
+    return self
+
+
+def load_midas_transform(model_type):
+    # https://github.com/isl-org/MiDaS/blob/master/run.py
+    # load transform only
+    if model_type == "dpt_large":  # DPT-Large
+        net_w, net_h = 384, 384
+        resize_mode = "minimal"
+        normalization = NormalizeImage(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
+
+    elif model_type == "dpt_hybrid":  # DPT-Hybrid
+        net_w, net_h = 384, 384
+        resize_mode = "minimal"
+        normalization = NormalizeImage(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
+
+    elif model_type == "midas_v21":
+        net_w, net_h = 384, 384
+        resize_mode = "upper_bound"
+        normalization = NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
+
+    elif model_type == "midas_v21_small":
+        net_w, net_h = 256, 256
+        resize_mode = "upper_bound"
+        normalization = NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
+
+    else:
+        assert False, f"model_type '{model_type}' not implemented, use: --model_type large"
+
+    transform = Compose(
+        [
+            Resize(
+                net_w,
+                net_h,
+                resize_target=None,
+                keep_aspect_ratio=True,
+                ensure_multiple_of=32,
+                resize_method=resize_mode,
+                image_interpolation_method=cv2.INTER_CUBIC,
+            ),
+            normalization,
+            PrepareForNet(),
+        ]
+    )
+
+    return transform
+
+
+def load_model(model_type):
+    # https://github.com/isl-org/MiDaS/blob/master/run.py
+    # load network
+    model_path = ISL_PATHS[model_type]
+    if model_type == "dpt_large":  # DPT-Large
+        model = DPTDepthModel(
+            path=model_path,
+            backbone="vitl16_384",
+            non_negative=True,
+        )
+        net_w, net_h = 384, 384
+        resize_mode = "minimal"
+        normalization = NormalizeImage(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
+
+    elif model_type == "dpt_hybrid":  # DPT-Hybrid
+        if not os.path.exists(model_path):
+            from basicsr.utils.download_util import load_file_from_url
+            load_file_from_url(remote_model_path, model_dir=annotator_ckpts_path)
+
+        model = DPTDepthModel(
+            path=model_path,
+            backbone="vitb_rn50_384",
+            non_negative=True,
+        )
+        net_w, net_h = 384, 384
+        resize_mode = "minimal"
+        normalization = NormalizeImage(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
+
+    elif model_type == "midas_v21":
+        model = MidasNet(model_path, non_negative=True)
+        net_w, net_h = 384, 384
+        resize_mode = "upper_bound"
+        normalization = NormalizeImage(
+            mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]
+        )
+
+    elif model_type == "midas_v21_small":
+        model = MidasNet_small(model_path, features=64, backbone="efficientnet_lite3", exportable=True,
+                               non_negative=True, blocks={'expand': True})
+        net_w, net_h = 256, 256
+        resize_mode = "upper_bound"
+        normalization = NormalizeImage(
+            mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]
+        )
+
+    else:
+        print(f"model_type '{model_type}' not implemented, use: --model_type large")
+        assert False
+
+    transform = Compose(
+        [
+            Resize(
+                net_w,
+                net_h,
+                resize_target=None,
+                keep_aspect_ratio=True,
+                ensure_multiple_of=32,
+                resize_method=resize_mode,
+                image_interpolation_method=cv2.INTER_CUBIC,
+            ),
+            normalization,
+            PrepareForNet(),
+        ]
+    )
+
+    return model.eval(), transform
+
+
+class MiDaSInference(nn.Module):
+    MODEL_TYPES_TORCH_HUB = [
+        "DPT_Large",
+        "DPT_Hybrid",
+        "MiDaS_small"
+    ]
+    MODEL_TYPES_ISL = [
+        "dpt_large",
+        "dpt_hybrid",
+        "midas_v21",
+        "midas_v21_small",
+    ]
+
+    def __init__(self, model_type):
+        super().__init__()
+        assert (model_type in self.MODEL_TYPES_ISL)
+        model, _ = load_model(model_type)
+        self.model = model
+        self.model.train = disabled_train
+
+    def forward(self, x):
+        with torch.no_grad():
+            prediction = self.model(x)
+        return prediction
+
diff --git a/annotator/midas/midas/__init__.py b/annotator/midas/midas/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/annotator/midas/midas/base_model.py b/annotator/midas/midas/base_model.py
new file mode 100644
index 0000000000000000000000000000000000000000..5cf430239b47ec5ec07531263f26f5c24a2311cd
--- /dev/null
+++ b/annotator/midas/midas/base_model.py
@@ -0,0 +1,16 @@
+import torch
+
+
+class BaseModel(torch.nn.Module):
+    def load(self, path):
+        """Load model from file.
+
+        Args:
+            path (str): file path
+        """
+        parameters = torch.load(path, map_location=torch.device('cpu'))
+
+        if "optimizer" in parameters:
+            parameters = parameters["model"]
+
+        self.load_state_dict(parameters)
diff --git a/annotator/midas/midas/blocks.py b/annotator/midas/midas/blocks.py
new file mode 100644
index 0000000000000000000000000000000000000000..2145d18fa98060a618536d9a64fe6589e9be4f78
--- /dev/null
+++ b/annotator/midas/midas/blocks.py
@@ -0,0 +1,342 @@
+import torch
+import torch.nn as nn
+
+from .vit import (
+    _make_pretrained_vitb_rn50_384,
+    _make_pretrained_vitl16_384,
+    _make_pretrained_vitb16_384,
+    forward_vit,
+)
+
+def _make_encoder(backbone, features, use_pretrained, groups=1, expand=False, exportable=True, hooks=None, use_vit_only=False, use_readout="ignore",):
+    if backbone == "vitl16_384":
+        pretrained = _make_pretrained_vitl16_384(
+            use_pretrained, hooks=hooks, use_readout=use_readout
+        )
+        scratch = _make_scratch(
+            [256, 512, 1024, 1024], features, groups=groups, expand=expand
+        )  # ViT-L/16 - 85.0% Top1 (backbone)
+    elif backbone == "vitb_rn50_384":
+        pretrained = _make_pretrained_vitb_rn50_384(
+            use_pretrained,
+            hooks=hooks,
+            use_vit_only=use_vit_only,
+            use_readout=use_readout,
+        )
+        scratch = _make_scratch(
+            [256, 512, 768, 768], features, groups=groups, expand=expand
+        )  # ViT-H/16 - 85.0% Top1 (backbone)
+    elif backbone == "vitb16_384":
+        pretrained = _make_pretrained_vitb16_384(
+            use_pretrained, hooks=hooks, use_readout=use_readout
+        )
+        scratch = _make_scratch(
+            [96, 192, 384, 768], features, groups=groups, expand=expand
+        )  # ViT-B/16 - 84.6% Top1 (backbone)
+    elif backbone == "resnext101_wsl":
+        pretrained = _make_pretrained_resnext101_wsl(use_pretrained)
+        scratch = _make_scratch([256, 512, 1024, 2048], features, groups=groups, expand=expand)     # efficientnet_lite3  
+    elif backbone == "efficientnet_lite3":
+        pretrained = _make_pretrained_efficientnet_lite3(use_pretrained, exportable=exportable)
+        scratch = _make_scratch([32, 48, 136, 384], features, groups=groups, expand=expand)  # efficientnet_lite3     
+    else:
+        print(f"Backbone '{backbone}' not implemented")
+        assert False
+        
+    return pretrained, scratch
+
+
+def _make_scratch(in_shape, out_shape, groups=1, expand=False):
+    scratch = nn.Module()
+
+    out_shape1 = out_shape
+    out_shape2 = out_shape
+    out_shape3 = out_shape
+    out_shape4 = out_shape
+    if expand==True:
+        out_shape1 = out_shape
+        out_shape2 = out_shape*2
+        out_shape3 = out_shape*4
+        out_shape4 = out_shape*8
+
+    scratch.layer1_rn = nn.Conv2d(
+        in_shape[0], out_shape1, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
+    )
+    scratch.layer2_rn = nn.Conv2d(
+        in_shape[1], out_shape2, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
+    )
+    scratch.layer3_rn = nn.Conv2d(
+        in_shape[2], out_shape3, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
+    )
+    scratch.layer4_rn = nn.Conv2d(
+        in_shape[3], out_shape4, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
+    )
+
+    return scratch
+
+
+def _make_pretrained_efficientnet_lite3(use_pretrained, exportable=False):
+    efficientnet = torch.hub.load(
+        "rwightman/gen-efficientnet-pytorch",
+        "tf_efficientnet_lite3",
+        pretrained=use_pretrained,
+        exportable=exportable
+    )
+    return _make_efficientnet_backbone(efficientnet)
+
+
+def _make_efficientnet_backbone(effnet):
+    pretrained = nn.Module()
+
+    pretrained.layer1 = nn.Sequential(
+        effnet.conv_stem, effnet.bn1, effnet.act1, *effnet.blocks[0:2]
+    )
+    pretrained.layer2 = nn.Sequential(*effnet.blocks[2:3])
+    pretrained.layer3 = nn.Sequential(*effnet.blocks[3:5])
+    pretrained.layer4 = nn.Sequential(*effnet.blocks[5:9])
+
+    return pretrained
+    
+
+def _make_resnet_backbone(resnet):
+    pretrained = nn.Module()
+    pretrained.layer1 = nn.Sequential(
+        resnet.conv1, resnet.bn1, resnet.relu, resnet.maxpool, resnet.layer1
+    )
+
+    pretrained.layer2 = resnet.layer2
+    pretrained.layer3 = resnet.layer3
+    pretrained.layer4 = resnet.layer4
+
+    return pretrained
+
+
+def _make_pretrained_resnext101_wsl(use_pretrained):
+    resnet = torch.hub.load("facebookresearch/WSL-Images", "resnext101_32x8d_wsl")
+    return _make_resnet_backbone(resnet)
+
+
+
+class Interpolate(nn.Module):
+    """Interpolation module.
+    """
+
+    def __init__(self, scale_factor, mode, align_corners=False):
+        """Init.
+
+        Args:
+            scale_factor (float): scaling
+            mode (str): interpolation mode
+        """
+        super(Interpolate, self).__init__()
+
+        self.interp = nn.functional.interpolate
+        self.scale_factor = scale_factor
+        self.mode = mode
+        self.align_corners = align_corners
+
+    def forward(self, x):
+        """Forward pass.
+
+        Args:
+            x (tensor): input
+
+        Returns:
+            tensor: interpolated data
+        """
+
+        x = self.interp(
+            x, scale_factor=self.scale_factor, mode=self.mode, align_corners=self.align_corners
+        )
+
+        return x
+
+
+class ResidualConvUnit(nn.Module):
+    """Residual convolution module.
+    """
+
+    def __init__(self, features):
+        """Init.
+
+        Args:
+            features (int): number of features
+        """
+        super().__init__()
+
+        self.conv1 = nn.Conv2d(
+            features, features, kernel_size=3, stride=1, padding=1, bias=True
+        )
+
+        self.conv2 = nn.Conv2d(
+            features, features, kernel_size=3, stride=1, padding=1, bias=True
+        )
+
+        self.relu = nn.ReLU(inplace=True)
+
+    def forward(self, x):
+        """Forward pass.
+
+        Args:
+            x (tensor): input
+
+        Returns:
+            tensor: output
+        """
+        out = self.relu(x)
+        out = self.conv1(out)
+        out = self.relu(out)
+        out = self.conv2(out)
+
+        return out + x
+
+
+class FeatureFusionBlock(nn.Module):
+    """Feature fusion block.
+    """
+
+    def __init__(self, features):
+        """Init.
+
+        Args:
+            features (int): number of features
+        """
+        super(FeatureFusionBlock, self).__init__()
+
+        self.resConfUnit1 = ResidualConvUnit(features)
+        self.resConfUnit2 = ResidualConvUnit(features)
+
+    def forward(self, *xs):
+        """Forward pass.
+
+        Returns:
+            tensor: output
+        """
+        output = xs[0]
+
+        if len(xs) == 2:
+            output += self.resConfUnit1(xs[1])
+
+        output = self.resConfUnit2(output)
+
+        output = nn.functional.interpolate(
+            output, scale_factor=2, mode="bilinear", align_corners=True
+        )
+
+        return output
+
+
+
+
+class ResidualConvUnit_custom(nn.Module):
+    """Residual convolution module.
+    """
+
+    def __init__(self, features, activation, bn):
+        """Init.
+
+        Args:
+            features (int): number of features
+        """
+        super().__init__()
+
+        self.bn = bn
+
+        self.groups=1
+
+        self.conv1 = nn.Conv2d(
+            features, features, kernel_size=3, stride=1, padding=1, bias=True, groups=self.groups
+        )
+        
+        self.conv2 = nn.Conv2d(
+            features, features, kernel_size=3, stride=1, padding=1, bias=True, groups=self.groups
+        )
+
+        if self.bn==True:
+            self.bn1 = nn.BatchNorm2d(features)
+            self.bn2 = nn.BatchNorm2d(features)
+
+        self.activation = activation
+
+        self.skip_add = nn.quantized.FloatFunctional()
+
+    def forward(self, x):
+        """Forward pass.
+
+        Args:
+            x (tensor): input
+
+        Returns:
+            tensor: output
+        """
+        
+        out = self.activation(x)
+        out = self.conv1(out)
+        if self.bn==True:
+            out = self.bn1(out)
+       
+        out = self.activation(out)
+        out = self.conv2(out)
+        if self.bn==True:
+            out = self.bn2(out)
+
+        if self.groups > 1:
+            out = self.conv_merge(out)
+
+        return self.skip_add.add(out, x)
+
+        # return out + x
+
+
+class FeatureFusionBlock_custom(nn.Module):
+    """Feature fusion block.
+    """
+
+    def __init__(self, features, activation, deconv=False, bn=False, expand=False, align_corners=True):
+        """Init.
+
+        Args:
+            features (int): number of features
+        """
+        super(FeatureFusionBlock_custom, self).__init__()
+
+        self.deconv = deconv
+        self.align_corners = align_corners
+
+        self.groups=1
+
+        self.expand = expand
+        out_features = features
+        if self.expand==True:
+            out_features = features//2
+        
+        self.out_conv = nn.Conv2d(features, out_features, kernel_size=1, stride=1, padding=0, bias=True, groups=1)
+
+        self.resConfUnit1 = ResidualConvUnit_custom(features, activation, bn)
+        self.resConfUnit2 = ResidualConvUnit_custom(features, activation, bn)
+        
+        self.skip_add = nn.quantized.FloatFunctional()
+
+    def forward(self, *xs):
+        """Forward pass.
+
+        Returns:
+            tensor: output
+        """
+        output = xs[0]
+
+        if len(xs) == 2:
+            res = self.resConfUnit1(xs[1])
+            output = self.skip_add.add(output, res)
+            # output += res
+
+        output = self.resConfUnit2(output)
+
+        output = nn.functional.interpolate(
+            output, scale_factor=2, mode="bilinear", align_corners=self.align_corners
+        )
+
+        output = self.out_conv(output)
+
+        return output
+
diff --git a/annotator/midas/midas/dpt_depth.py b/annotator/midas/midas/dpt_depth.py
new file mode 100644
index 0000000000000000000000000000000000000000..4e9aab5d2767dffea39da5b3f30e2798688216f1
--- /dev/null
+++ b/annotator/midas/midas/dpt_depth.py
@@ -0,0 +1,109 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from .base_model import BaseModel
+from .blocks import (
+    FeatureFusionBlock,
+    FeatureFusionBlock_custom,
+    Interpolate,
+    _make_encoder,
+    forward_vit,
+)
+
+
+def _make_fusion_block(features, use_bn):
+    return FeatureFusionBlock_custom(
+        features,
+        nn.ReLU(False),
+        deconv=False,
+        bn=use_bn,
+        expand=False,
+        align_corners=True,
+    )
+
+
+class DPT(BaseModel):
+    def __init__(
+        self,
+        head,
+        features=256,
+        backbone="vitb_rn50_384",
+        readout="project",
+        channels_last=False,
+        use_bn=False,
+    ):
+
+        super(DPT, self).__init__()
+
+        self.channels_last = channels_last
+
+        hooks = {
+            "vitb_rn50_384": [0, 1, 8, 11],
+            "vitb16_384": [2, 5, 8, 11],
+            "vitl16_384": [5, 11, 17, 23],
+        }
+
+        # Instantiate backbone and reassemble blocks
+        self.pretrained, self.scratch = _make_encoder(
+            backbone,
+            features,
+            False, # Set to true of you want to train from scratch, uses ImageNet weights
+            groups=1,
+            expand=False,
+            exportable=False,
+            hooks=hooks[backbone],
+            use_readout=readout,
+        )
+
+        self.scratch.refinenet1 = _make_fusion_block(features, use_bn)
+        self.scratch.refinenet2 = _make_fusion_block(features, use_bn)
+        self.scratch.refinenet3 = _make_fusion_block(features, use_bn)
+        self.scratch.refinenet4 = _make_fusion_block(features, use_bn)
+
+        self.scratch.output_conv = head
+
+
+    def forward(self, x):
+        if self.channels_last == True:
+            x.contiguous(memory_format=torch.channels_last)
+
+        layer_1, layer_2, layer_3, layer_4 = forward_vit(self.pretrained, x)
+
+        layer_1_rn = self.scratch.layer1_rn(layer_1)
+        layer_2_rn = self.scratch.layer2_rn(layer_2)
+        layer_3_rn = self.scratch.layer3_rn(layer_3)
+        layer_4_rn = self.scratch.layer4_rn(layer_4)
+
+        path_4 = self.scratch.refinenet4(layer_4_rn)
+        path_3 = self.scratch.refinenet3(path_4, layer_3_rn)
+        path_2 = self.scratch.refinenet2(path_3, layer_2_rn)
+        path_1 = self.scratch.refinenet1(path_2, layer_1_rn)
+
+        out = self.scratch.output_conv(path_1)
+
+        return out
+
+
+class DPTDepthModel(DPT):
+    def __init__(self, path=None, non_negative=True, **kwargs):
+        features = kwargs["features"] if "features" in kwargs else 256
+
+        head = nn.Sequential(
+            nn.Conv2d(features, features // 2, kernel_size=3, stride=1, padding=1),
+            Interpolate(scale_factor=2, mode="bilinear", align_corners=True),
+            nn.Conv2d(features // 2, 32, kernel_size=3, stride=1, padding=1),
+            nn.ReLU(True),
+            nn.Conv2d(32, 1, kernel_size=1, stride=1, padding=0),
+            nn.ReLU(True) if non_negative else nn.Identity(),
+            nn.Identity(),
+        )
+
+        super().__init__(head, **kwargs)
+
+        if path is not None:
+           self.load(path)
+
+    def forward(self, x):
+        return super().forward(x).squeeze(dim=1)
+
diff --git a/annotator/midas/midas/midas_net.py b/annotator/midas/midas/midas_net.py
new file mode 100644
index 0000000000000000000000000000000000000000..8a954977800b0a0f48807e80fa63041910e33c1f
--- /dev/null
+++ b/annotator/midas/midas/midas_net.py
@@ -0,0 +1,76 @@
+"""MidashNet: Network for monocular depth estimation trained by mixing several datasets.
+This file contains code that is adapted from
+https://github.com/thomasjpfan/pytorch_refinenet/blob/master/pytorch_refinenet/refinenet/refinenet_4cascade.py
+"""
+import torch
+import torch.nn as nn
+
+from .base_model import BaseModel
+from .blocks import FeatureFusionBlock, Interpolate, _make_encoder
+
+
+class MidasNet(BaseModel):
+    """Network for monocular depth estimation.
+    """
+
+    def __init__(self, path=None, features=256, non_negative=True):
+        """Init.
+
+        Args:
+            path (str, optional): Path to saved model. Defaults to None.
+            features (int, optional): Number of features. Defaults to 256.
+            backbone (str, optional): Backbone network for encoder. Defaults to resnet50
+        """
+        print("Loading weights: ", path)
+
+        super(MidasNet, self).__init__()
+
+        use_pretrained = False if path is None else True
+
+        self.pretrained, self.scratch = _make_encoder(backbone="resnext101_wsl", features=features, use_pretrained=use_pretrained)
+
+        self.scratch.refinenet4 = FeatureFusionBlock(features)
+        self.scratch.refinenet3 = FeatureFusionBlock(features)
+        self.scratch.refinenet2 = FeatureFusionBlock(features)
+        self.scratch.refinenet1 = FeatureFusionBlock(features)
+
+        self.scratch.output_conv = nn.Sequential(
+            nn.Conv2d(features, 128, kernel_size=3, stride=1, padding=1),
+            Interpolate(scale_factor=2, mode="bilinear"),
+            nn.Conv2d(128, 32, kernel_size=3, stride=1, padding=1),
+            nn.ReLU(True),
+            nn.Conv2d(32, 1, kernel_size=1, stride=1, padding=0),
+            nn.ReLU(True) if non_negative else nn.Identity(),
+        )
+
+        if path:
+            self.load(path)
+
+    def forward(self, x):
+        """Forward pass.
+
+        Args:
+            x (tensor): input data (image)
+
+        Returns:
+            tensor: depth
+        """
+
+        layer_1 = self.pretrained.layer1(x)
+        layer_2 = self.pretrained.layer2(layer_1)
+        layer_3 = self.pretrained.layer3(layer_2)
+        layer_4 = self.pretrained.layer4(layer_3)
+
+        layer_1_rn = self.scratch.layer1_rn(layer_1)
+        layer_2_rn = self.scratch.layer2_rn(layer_2)
+        layer_3_rn = self.scratch.layer3_rn(layer_3)
+        layer_4_rn = self.scratch.layer4_rn(layer_4)
+
+        path_4 = self.scratch.refinenet4(layer_4_rn)
+        path_3 = self.scratch.refinenet3(path_4, layer_3_rn)
+        path_2 = self.scratch.refinenet2(path_3, layer_2_rn)
+        path_1 = self.scratch.refinenet1(path_2, layer_1_rn)
+
+        out = self.scratch.output_conv(path_1)
+
+        return torch.squeeze(out, dim=1)
diff --git a/annotator/midas/midas/midas_net_custom.py b/annotator/midas/midas/midas_net_custom.py
new file mode 100644
index 0000000000000000000000000000000000000000..50e4acb5e53d5fabefe3dde16ab49c33c2b7797c
--- /dev/null
+++ b/annotator/midas/midas/midas_net_custom.py
@@ -0,0 +1,128 @@
+"""MidashNet: Network for monocular depth estimation trained by mixing several datasets.
+This file contains code that is adapted from
+https://github.com/thomasjpfan/pytorch_refinenet/blob/master/pytorch_refinenet/refinenet/refinenet_4cascade.py
+"""
+import torch
+import torch.nn as nn
+
+from .base_model import BaseModel
+from .blocks import FeatureFusionBlock, FeatureFusionBlock_custom, Interpolate, _make_encoder
+
+
+class MidasNet_small(BaseModel):
+    """Network for monocular depth estimation.
+    """
+
+    def __init__(self, path=None, features=64, backbone="efficientnet_lite3", non_negative=True, exportable=True, channels_last=False, align_corners=True,
+        blocks={'expand': True}):
+        """Init.
+
+        Args:
+            path (str, optional): Path to saved model. Defaults to None.
+            features (int, optional): Number of features. Defaults to 256.
+            backbone (str, optional): Backbone network for encoder. Defaults to resnet50
+        """
+        print("Loading weights: ", path)
+
+        super(MidasNet_small, self).__init__()
+
+        use_pretrained = False if path else True
+                
+        self.channels_last = channels_last
+        self.blocks = blocks
+        self.backbone = backbone
+
+        self.groups = 1
+
+        features1=features
+        features2=features
+        features3=features
+        features4=features
+        self.expand = False
+        if "expand" in self.blocks and self.blocks['expand'] == True:
+            self.expand = True
+            features1=features
+            features2=features*2
+            features3=features*4
+            features4=features*8
+
+        self.pretrained, self.scratch = _make_encoder(self.backbone, features, use_pretrained, groups=self.groups, expand=self.expand, exportable=exportable)
+  
+        self.scratch.activation = nn.ReLU(False)    
+
+        self.scratch.refinenet4 = FeatureFusionBlock_custom(features4, self.scratch.activation, deconv=False, bn=False, expand=self.expand, align_corners=align_corners)
+        self.scratch.refinenet3 = FeatureFusionBlock_custom(features3, self.scratch.activation, deconv=False, bn=False, expand=self.expand, align_corners=align_corners)
+        self.scratch.refinenet2 = FeatureFusionBlock_custom(features2, self.scratch.activation, deconv=False, bn=False, expand=self.expand, align_corners=align_corners)
+        self.scratch.refinenet1 = FeatureFusionBlock_custom(features1, self.scratch.activation, deconv=False, bn=False, align_corners=align_corners)
+
+        
+        self.scratch.output_conv = nn.Sequential(
+            nn.Conv2d(features, features//2, kernel_size=3, stride=1, padding=1, groups=self.groups),
+            Interpolate(scale_factor=2, mode="bilinear"),
+            nn.Conv2d(features//2, 32, kernel_size=3, stride=1, padding=1),
+            self.scratch.activation,
+            nn.Conv2d(32, 1, kernel_size=1, stride=1, padding=0),
+            nn.ReLU(True) if non_negative else nn.Identity(),
+            nn.Identity(),
+        )
+        
+        if path:
+            self.load(path)
+
+
+    def forward(self, x):
+        """Forward pass.
+
+        Args:
+            x (tensor): input data (image)
+
+        Returns:
+            tensor: depth
+        """
+        if self.channels_last==True:
+            print("self.channels_last = ", self.channels_last)
+            x.contiguous(memory_format=torch.channels_last)
+
+
+        layer_1 = self.pretrained.layer1(x)
+        layer_2 = self.pretrained.layer2(layer_1)
+        layer_3 = self.pretrained.layer3(layer_2)
+        layer_4 = self.pretrained.layer4(layer_3)
+        
+        layer_1_rn = self.scratch.layer1_rn(layer_1)
+        layer_2_rn = self.scratch.layer2_rn(layer_2)
+        layer_3_rn = self.scratch.layer3_rn(layer_3)
+        layer_4_rn = self.scratch.layer4_rn(layer_4)
+
+
+        path_4 = self.scratch.refinenet4(layer_4_rn)
+        path_3 = self.scratch.refinenet3(path_4, layer_3_rn)
+        path_2 = self.scratch.refinenet2(path_3, layer_2_rn)
+        path_1 = self.scratch.refinenet1(path_2, layer_1_rn)
+        
+        out = self.scratch.output_conv(path_1)
+
+        return torch.squeeze(out, dim=1)
+
+
+
+def fuse_model(m):
+    prev_previous_type = nn.Identity()
+    prev_previous_name = ''
+    previous_type = nn.Identity()
+    previous_name = ''
+    for name, module in m.named_modules():
+        if prev_previous_type == nn.Conv2d and previous_type == nn.BatchNorm2d and type(module) == nn.ReLU:
+            # print("FUSED ", prev_previous_name, previous_name, name)
+            torch.quantization.fuse_modules(m, [prev_previous_name, previous_name, name], inplace=True)
+        elif prev_previous_type == nn.Conv2d and previous_type == nn.BatchNorm2d:
+            # print("FUSED ", prev_previous_name, previous_name)
+            torch.quantization.fuse_modules(m, [prev_previous_name, previous_name], inplace=True)
+        # elif previous_type == nn.Conv2d and type(module) == nn.ReLU:
+        #    print("FUSED ", previous_name, name)
+        #    torch.quantization.fuse_modules(m, [previous_name, name], inplace=True)
+
+        prev_previous_type = previous_type
+        prev_previous_name = previous_name
+        previous_type = type(module)
+        previous_name = name
\ No newline at end of file
diff --git a/annotator/midas/midas/transforms.py b/annotator/midas/midas/transforms.py
new file mode 100644
index 0000000000000000000000000000000000000000..350cbc11662633ad7f8968eb10be2e7de6e384e9
--- /dev/null
+++ b/annotator/midas/midas/transforms.py
@@ -0,0 +1,234 @@
+import numpy as np
+import cv2
+import math
+
+
+def apply_min_size(sample, size, image_interpolation_method=cv2.INTER_AREA):
+    """Rezise the sample to ensure the given size. Keeps aspect ratio.
+
+    Args:
+        sample (dict): sample
+        size (tuple): image size
+
+    Returns:
+        tuple: new size
+    """
+    shape = list(sample["disparity"].shape)
+
+    if shape[0] >= size[0] and shape[1] >= size[1]:
+        return sample
+
+    scale = [0, 0]
+    scale[0] = size[0] / shape[0]
+    scale[1] = size[1] / shape[1]
+
+    scale = max(scale)
+
+    shape[0] = math.ceil(scale * shape[0])
+    shape[1] = math.ceil(scale * shape[1])
+
+    # resize
+    sample["image"] = cv2.resize(
+        sample["image"], tuple(shape[::-1]), interpolation=image_interpolation_method
+    )
+
+    sample["disparity"] = cv2.resize(
+        sample["disparity"], tuple(shape[::-1]), interpolation=cv2.INTER_NEAREST
+    )
+    sample["mask"] = cv2.resize(
+        sample["mask"].astype(np.float32),
+        tuple(shape[::-1]),
+        interpolation=cv2.INTER_NEAREST,
+    )
+    sample["mask"] = sample["mask"].astype(bool)
+
+    return tuple(shape)
+
+
+class Resize(object):
+    """Resize sample to given size (width, height).
+    """
+
+    def __init__(
+        self,
+        width,
+        height,
+        resize_target=True,
+        keep_aspect_ratio=False,
+        ensure_multiple_of=1,
+        resize_method="lower_bound",
+        image_interpolation_method=cv2.INTER_AREA,
+    ):
+        """Init.
+
+        Args:
+            width (int): desired output width
+            height (int): desired output height
+            resize_target (bool, optional):
+                True: Resize the full sample (image, mask, target).
+                False: Resize image only.
+                Defaults to True.
+            keep_aspect_ratio (bool, optional):
+                True: Keep the aspect ratio of the input sample.
+                Output sample might not have the given width and height, and
+                resize behaviour depends on the parameter 'resize_method'.
+                Defaults to False.
+            ensure_multiple_of (int, optional):
+                Output width and height is constrained to be multiple of this parameter.
+                Defaults to 1.
+            resize_method (str, optional):
+                "lower_bound": Output will be at least as large as the given size.
+                "upper_bound": Output will be at max as large as the given size. (Output size might be smaller than given size.)
+                "minimal": Scale as least as possible.  (Output size might be smaller than given size.)
+                Defaults to "lower_bound".
+        """
+        self.__width = width
+        self.__height = height
+
+        self.__resize_target = resize_target
+        self.__keep_aspect_ratio = keep_aspect_ratio
+        self.__multiple_of = ensure_multiple_of
+        self.__resize_method = resize_method
+        self.__image_interpolation_method = image_interpolation_method
+
+    def constrain_to_multiple_of(self, x, min_val=0, max_val=None):
+        y = (np.round(x / self.__multiple_of) * self.__multiple_of).astype(int)
+
+        if max_val is not None and y > max_val:
+            y = (np.floor(x / self.__multiple_of) * self.__multiple_of).astype(int)
+
+        if y < min_val:
+            y = (np.ceil(x / self.__multiple_of) * self.__multiple_of).astype(int)
+
+        return y
+
+    def get_size(self, width, height):
+        # determine new height and width
+        scale_height = self.__height / height
+        scale_width = self.__width / width
+
+        if self.__keep_aspect_ratio:
+            if self.__resize_method == "lower_bound":
+                # scale such that output size is lower bound
+                if scale_width > scale_height:
+                    # fit width
+                    scale_height = scale_width
+                else:
+                    # fit height
+                    scale_width = scale_height
+            elif self.__resize_method == "upper_bound":
+                # scale such that output size is upper bound
+                if scale_width < scale_height:
+                    # fit width
+                    scale_height = scale_width
+                else:
+                    # fit height
+                    scale_width = scale_height
+            elif self.__resize_method == "minimal":
+                # scale as least as possbile
+                if abs(1 - scale_width) < abs(1 - scale_height):
+                    # fit width
+                    scale_height = scale_width
+                else:
+                    # fit height
+                    scale_width = scale_height
+            else:
+                raise ValueError(
+                    f"resize_method {self.__resize_method} not implemented"
+                )
+
+        if self.__resize_method == "lower_bound":
+            new_height = self.constrain_to_multiple_of(
+                scale_height * height, min_val=self.__height
+            )
+            new_width = self.constrain_to_multiple_of(
+                scale_width * width, min_val=self.__width
+            )
+        elif self.__resize_method == "upper_bound":
+            new_height = self.constrain_to_multiple_of(
+                scale_height * height, max_val=self.__height
+            )
+            new_width = self.constrain_to_multiple_of(
+                scale_width * width, max_val=self.__width
+            )
+        elif self.__resize_method == "minimal":
+            new_height = self.constrain_to_multiple_of(scale_height * height)
+            new_width = self.constrain_to_multiple_of(scale_width * width)
+        else:
+            raise ValueError(f"resize_method {self.__resize_method} not implemented")
+
+        return (new_width, new_height)
+
+    def __call__(self, sample):
+        width, height = self.get_size(
+            sample["image"].shape[1], sample["image"].shape[0]
+        )
+
+        # resize sample
+        sample["image"] = cv2.resize(
+            sample["image"],
+            (width, height),
+            interpolation=self.__image_interpolation_method,
+        )
+
+        if self.__resize_target:
+            if "disparity" in sample:
+                sample["disparity"] = cv2.resize(
+                    sample["disparity"],
+                    (width, height),
+                    interpolation=cv2.INTER_NEAREST,
+                )
+
+            if "depth" in sample:
+                sample["depth"] = cv2.resize(
+                    sample["depth"], (width, height), interpolation=cv2.INTER_NEAREST
+                )
+
+            sample["mask"] = cv2.resize(
+                sample["mask"].astype(np.float32),
+                (width, height),
+                interpolation=cv2.INTER_NEAREST,
+            )
+            sample["mask"] = sample["mask"].astype(bool)
+
+        return sample
+
+
+class NormalizeImage(object):
+    """Normlize image by given mean and std.
+    """
+
+    def __init__(self, mean, std):
+        self.__mean = mean
+        self.__std = std
+
+    def __call__(self, sample):
+        sample["image"] = (sample["image"] - self.__mean) / self.__std
+
+        return sample
+
+
+class PrepareForNet(object):
+    """Prepare sample for usage as network input.
+    """
+
+    def __init__(self):
+        pass
+
+    def __call__(self, sample):
+        image = np.transpose(sample["image"], (2, 0, 1))
+        sample["image"] = np.ascontiguousarray(image).astype(np.float32)
+
+        if "mask" in sample:
+            sample["mask"] = sample["mask"].astype(np.float32)
+            sample["mask"] = np.ascontiguousarray(sample["mask"])
+
+        if "disparity" in sample:
+            disparity = sample["disparity"].astype(np.float32)
+            sample["disparity"] = np.ascontiguousarray(disparity)
+
+        if "depth" in sample:
+            depth = sample["depth"].astype(np.float32)
+            sample["depth"] = np.ascontiguousarray(depth)
+
+        return sample
diff --git a/annotator/midas/midas/vit.py b/annotator/midas/midas/vit.py
new file mode 100644
index 0000000000000000000000000000000000000000..ea46b1be88b261b0dec04f3da0256f5f66f88a74
--- /dev/null
+++ b/annotator/midas/midas/vit.py
@@ -0,0 +1,491 @@
+import torch
+import torch.nn as nn
+import timm
+import types
+import math
+import torch.nn.functional as F
+
+
+class Slice(nn.Module):
+    def __init__(self, start_index=1):
+        super(Slice, self).__init__()
+        self.start_index = start_index
+
+    def forward(self, x):
+        return x[:, self.start_index :]
+
+
+class AddReadout(nn.Module):
+    def __init__(self, start_index=1):
+        super(AddReadout, self).__init__()
+        self.start_index = start_index
+
+    def forward(self, x):
+        if self.start_index == 2:
+            readout = (x[:, 0] + x[:, 1]) / 2
+        else:
+            readout = x[:, 0]
+        return x[:, self.start_index :] + readout.unsqueeze(1)
+
+
+class ProjectReadout(nn.Module):
+    def __init__(self, in_features, start_index=1):
+        super(ProjectReadout, self).__init__()
+        self.start_index = start_index
+
+        self.project = nn.Sequential(nn.Linear(2 * in_features, in_features), nn.GELU())
+
+    def forward(self, x):
+        readout = x[:, 0].unsqueeze(1).expand_as(x[:, self.start_index :])
+        features = torch.cat((x[:, self.start_index :], readout), -1)
+
+        return self.project(features)
+
+
+class Transpose(nn.Module):
+    def __init__(self, dim0, dim1):
+        super(Transpose, self).__init__()
+        self.dim0 = dim0
+        self.dim1 = dim1
+
+    def forward(self, x):
+        x = x.transpose(self.dim0, self.dim1)
+        return x
+
+
+def forward_vit(pretrained, x):
+    b, c, h, w = x.shape
+
+    glob = pretrained.model.forward_flex(x)
+
+    layer_1 = pretrained.activations["1"]
+    layer_2 = pretrained.activations["2"]
+    layer_3 = pretrained.activations["3"]
+    layer_4 = pretrained.activations["4"]
+
+    layer_1 = pretrained.act_postprocess1[0:2](layer_1)
+    layer_2 = pretrained.act_postprocess2[0:2](layer_2)
+    layer_3 = pretrained.act_postprocess3[0:2](layer_3)
+    layer_4 = pretrained.act_postprocess4[0:2](layer_4)
+
+    unflatten = nn.Sequential(
+        nn.Unflatten(
+            2,
+            torch.Size(
+                [
+                    h // pretrained.model.patch_size[1],
+                    w // pretrained.model.patch_size[0],
+                ]
+            ),
+        )
+    )
+
+    if layer_1.ndim == 3:
+        layer_1 = unflatten(layer_1)
+    if layer_2.ndim == 3:
+        layer_2 = unflatten(layer_2)
+    if layer_3.ndim == 3:
+        layer_3 = unflatten(layer_3)
+    if layer_4.ndim == 3:
+        layer_4 = unflatten(layer_4)
+
+    layer_1 = pretrained.act_postprocess1[3 : len(pretrained.act_postprocess1)](layer_1)
+    layer_2 = pretrained.act_postprocess2[3 : len(pretrained.act_postprocess2)](layer_2)
+    layer_3 = pretrained.act_postprocess3[3 : len(pretrained.act_postprocess3)](layer_3)
+    layer_4 = pretrained.act_postprocess4[3 : len(pretrained.act_postprocess4)](layer_4)
+
+    return layer_1, layer_2, layer_3, layer_4
+
+
+def _resize_pos_embed(self, posemb, gs_h, gs_w):
+    posemb_tok, posemb_grid = (
+        posemb[:, : self.start_index],
+        posemb[0, self.start_index :],
+    )
+
+    gs_old = int(math.sqrt(len(posemb_grid)))
+
+    posemb_grid = posemb_grid.reshape(1, gs_old, gs_old, -1).permute(0, 3, 1, 2)
+    posemb_grid = F.interpolate(posemb_grid, size=(gs_h, gs_w), mode="bilinear")
+    posemb_grid = posemb_grid.permute(0, 2, 3, 1).reshape(1, gs_h * gs_w, -1)
+
+    posemb = torch.cat([posemb_tok, posemb_grid], dim=1)
+
+    return posemb
+
+
+def forward_flex(self, x):
+    b, c, h, w = x.shape
+
+    pos_embed = self._resize_pos_embed(
+        self.pos_embed, h // self.patch_size[1], w // self.patch_size[0]
+    )
+
+    B = x.shape[0]
+
+    if hasattr(self.patch_embed, "backbone"):
+        x = self.patch_embed.backbone(x)
+        if isinstance(x, (list, tuple)):
+            x = x[-1]  # last feature if backbone outputs list/tuple of features
+
+    x = self.patch_embed.proj(x).flatten(2).transpose(1, 2)
+
+    if getattr(self, "dist_token", None) is not None:
+        cls_tokens = self.cls_token.expand(
+            B, -1, -1
+        )  # stole cls_tokens impl from Phil Wang, thanks
+        dist_token = self.dist_token.expand(B, -1, -1)
+        x = torch.cat((cls_tokens, dist_token, x), dim=1)
+    else:
+        cls_tokens = self.cls_token.expand(
+            B, -1, -1
+        )  # stole cls_tokens impl from Phil Wang, thanks
+        x = torch.cat((cls_tokens, x), dim=1)
+
+    x = x + pos_embed
+    x = self.pos_drop(x)
+
+    for blk in self.blocks:
+        x = blk(x)
+
+    x = self.norm(x)
+
+    return x
+
+
+activations = {}
+
+
+def get_activation(name):
+    def hook(model, input, output):
+        activations[name] = output
+
+    return hook
+
+
+def get_readout_oper(vit_features, features, use_readout, start_index=1):
+    if use_readout == "ignore":
+        readout_oper = [Slice(start_index)] * len(features)
+    elif use_readout == "add":
+        readout_oper = [AddReadout(start_index)] * len(features)
+    elif use_readout == "project":
+        readout_oper = [
+            ProjectReadout(vit_features, start_index) for out_feat in features
+        ]
+    else:
+        assert (
+            False
+        ), "wrong operation for readout token, use_readout can be 'ignore', 'add', or 'project'"
+
+    return readout_oper
+
+
+def _make_vit_b16_backbone(
+    model,
+    features=[96, 192, 384, 768],
+    size=[384, 384],
+    hooks=[2, 5, 8, 11],
+    vit_features=768,
+    use_readout="ignore",
+    start_index=1,
+):
+    pretrained = nn.Module()
+
+    pretrained.model = model
+    pretrained.model.blocks[hooks[0]].register_forward_hook(get_activation("1"))
+    pretrained.model.blocks[hooks[1]].register_forward_hook(get_activation("2"))
+    pretrained.model.blocks[hooks[2]].register_forward_hook(get_activation("3"))
+    pretrained.model.blocks[hooks[3]].register_forward_hook(get_activation("4"))
+
+    pretrained.activations = activations
+
+    readout_oper = get_readout_oper(vit_features, features, use_readout, start_index)
+
+    # 32, 48, 136, 384
+    pretrained.act_postprocess1 = nn.Sequential(
+        readout_oper[0],
+        Transpose(1, 2),
+        nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+        nn.Conv2d(
+            in_channels=vit_features,
+            out_channels=features[0],
+            kernel_size=1,
+            stride=1,
+            padding=0,
+        ),
+        nn.ConvTranspose2d(
+            in_channels=features[0],
+            out_channels=features[0],
+            kernel_size=4,
+            stride=4,
+            padding=0,
+            bias=True,
+            dilation=1,
+            groups=1,
+        ),
+    )
+
+    pretrained.act_postprocess2 = nn.Sequential(
+        readout_oper[1],
+        Transpose(1, 2),
+        nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+        nn.Conv2d(
+            in_channels=vit_features,
+            out_channels=features[1],
+            kernel_size=1,
+            stride=1,
+            padding=0,
+        ),
+        nn.ConvTranspose2d(
+            in_channels=features[1],
+            out_channels=features[1],
+            kernel_size=2,
+            stride=2,
+            padding=0,
+            bias=True,
+            dilation=1,
+            groups=1,
+        ),
+    )
+
+    pretrained.act_postprocess3 = nn.Sequential(
+        readout_oper[2],
+        Transpose(1, 2),
+        nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+        nn.Conv2d(
+            in_channels=vit_features,
+            out_channels=features[2],
+            kernel_size=1,
+            stride=1,
+            padding=0,
+        ),
+    )
+
+    pretrained.act_postprocess4 = nn.Sequential(
+        readout_oper[3],
+        Transpose(1, 2),
+        nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+        nn.Conv2d(
+            in_channels=vit_features,
+            out_channels=features[3],
+            kernel_size=1,
+            stride=1,
+            padding=0,
+        ),
+        nn.Conv2d(
+            in_channels=features[3],
+            out_channels=features[3],
+            kernel_size=3,
+            stride=2,
+            padding=1,
+        ),
+    )
+
+    pretrained.model.start_index = start_index
+    pretrained.model.patch_size = [16, 16]
+
+    # We inject this function into the VisionTransformer instances so that
+    # we can use it with interpolated position embeddings without modifying the library source.
+    pretrained.model.forward_flex = types.MethodType(forward_flex, pretrained.model)
+    pretrained.model._resize_pos_embed = types.MethodType(
+        _resize_pos_embed, pretrained.model
+    )
+
+    return pretrained
+
+
+def _make_pretrained_vitl16_384(pretrained, use_readout="ignore", hooks=None):
+    model = timm.create_model("vit_large_patch16_384", pretrained=pretrained)
+
+    hooks = [5, 11, 17, 23] if hooks == None else hooks
+    return _make_vit_b16_backbone(
+        model,
+        features=[256, 512, 1024, 1024],
+        hooks=hooks,
+        vit_features=1024,
+        use_readout=use_readout,
+    )
+
+
+def _make_pretrained_vitb16_384(pretrained, use_readout="ignore", hooks=None):
+    model = timm.create_model("vit_base_patch16_384", pretrained=pretrained)
+
+    hooks = [2, 5, 8, 11] if hooks == None else hooks
+    return _make_vit_b16_backbone(
+        model, features=[96, 192, 384, 768], hooks=hooks, use_readout=use_readout
+    )
+
+
+def _make_pretrained_deitb16_384(pretrained, use_readout="ignore", hooks=None):
+    model = timm.create_model("vit_deit_base_patch16_384", pretrained=pretrained)
+
+    hooks = [2, 5, 8, 11] if hooks == None else hooks
+    return _make_vit_b16_backbone(
+        model, features=[96, 192, 384, 768], hooks=hooks, use_readout=use_readout
+    )
+
+
+def _make_pretrained_deitb16_distil_384(pretrained, use_readout="ignore", hooks=None):
+    model = timm.create_model(
+        "vit_deit_base_distilled_patch16_384", pretrained=pretrained
+    )
+
+    hooks = [2, 5, 8, 11] if hooks == None else hooks
+    return _make_vit_b16_backbone(
+        model,
+        features=[96, 192, 384, 768],
+        hooks=hooks,
+        use_readout=use_readout,
+        start_index=2,
+    )
+
+
+def _make_vit_b_rn50_backbone(
+    model,
+    features=[256, 512, 768, 768],
+    size=[384, 384],
+    hooks=[0, 1, 8, 11],
+    vit_features=768,
+    use_vit_only=False,
+    use_readout="ignore",
+    start_index=1,
+):
+    pretrained = nn.Module()
+
+    pretrained.model = model
+
+    if use_vit_only == True:
+        pretrained.model.blocks[hooks[0]].register_forward_hook(get_activation("1"))
+        pretrained.model.blocks[hooks[1]].register_forward_hook(get_activation("2"))
+    else:
+        pretrained.model.patch_embed.backbone.stages[0].register_forward_hook(
+            get_activation("1")
+        )
+        pretrained.model.patch_embed.backbone.stages[1].register_forward_hook(
+            get_activation("2")
+        )
+
+    pretrained.model.blocks[hooks[2]].register_forward_hook(get_activation("3"))
+    pretrained.model.blocks[hooks[3]].register_forward_hook(get_activation("4"))
+
+    pretrained.activations = activations
+
+    readout_oper = get_readout_oper(vit_features, features, use_readout, start_index)
+
+    if use_vit_only == True:
+        pretrained.act_postprocess1 = nn.Sequential(
+            readout_oper[0],
+            Transpose(1, 2),
+            nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+            nn.Conv2d(
+                in_channels=vit_features,
+                out_channels=features[0],
+                kernel_size=1,
+                stride=1,
+                padding=0,
+            ),
+            nn.ConvTranspose2d(
+                in_channels=features[0],
+                out_channels=features[0],
+                kernel_size=4,
+                stride=4,
+                padding=0,
+                bias=True,
+                dilation=1,
+                groups=1,
+            ),
+        )
+
+        pretrained.act_postprocess2 = nn.Sequential(
+            readout_oper[1],
+            Transpose(1, 2),
+            nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+            nn.Conv2d(
+                in_channels=vit_features,
+                out_channels=features[1],
+                kernel_size=1,
+                stride=1,
+                padding=0,
+            ),
+            nn.ConvTranspose2d(
+                in_channels=features[1],
+                out_channels=features[1],
+                kernel_size=2,
+                stride=2,
+                padding=0,
+                bias=True,
+                dilation=1,
+                groups=1,
+            ),
+        )
+    else:
+        pretrained.act_postprocess1 = nn.Sequential(
+            nn.Identity(), nn.Identity(), nn.Identity()
+        )
+        pretrained.act_postprocess2 = nn.Sequential(
+            nn.Identity(), nn.Identity(), nn.Identity()
+        )
+
+    pretrained.act_postprocess3 = nn.Sequential(
+        readout_oper[2],
+        Transpose(1, 2),
+        nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+        nn.Conv2d(
+            in_channels=vit_features,
+            out_channels=features[2],
+            kernel_size=1,
+            stride=1,
+            padding=0,
+        ),
+    )
+
+    pretrained.act_postprocess4 = nn.Sequential(
+        readout_oper[3],
+        Transpose(1, 2),
+        nn.Unflatten(2, torch.Size([size[0] // 16, size[1] // 16])),
+        nn.Conv2d(
+            in_channels=vit_features,
+            out_channels=features[3],
+            kernel_size=1,
+            stride=1,
+            padding=0,
+        ),
+        nn.Conv2d(
+            in_channels=features[3],
+            out_channels=features[3],
+            kernel_size=3,
+            stride=2,
+            padding=1,
+        ),
+    )
+
+    pretrained.model.start_index = start_index
+    pretrained.model.patch_size = [16, 16]
+
+    # We inject this function into the VisionTransformer instances so that
+    # we can use it with interpolated position embeddings without modifying the library source.
+    pretrained.model.forward_flex = types.MethodType(forward_flex, pretrained.model)
+
+    # We inject this function into the VisionTransformer instances so that
+    # we can use it with interpolated position embeddings without modifying the library source.
+    pretrained.model._resize_pos_embed = types.MethodType(
+        _resize_pos_embed, pretrained.model
+    )
+
+    return pretrained
+
+
+def _make_pretrained_vitb_rn50_384(
+    pretrained, use_readout="ignore", hooks=None, use_vit_only=False
+):
+    model = timm.create_model("vit_base_resnet50_384", pretrained=pretrained)
+
+    hooks = [0, 1, 8, 11] if hooks == None else hooks
+    return _make_vit_b_rn50_backbone(
+        model,
+        features=[256, 512, 768, 768],
+        size=[384, 384],
+        hooks=hooks,
+        use_vit_only=use_vit_only,
+        use_readout=use_readout,
+    )
diff --git a/annotator/midas/utils.py b/annotator/midas/utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..9a9d3b5b66370fa98da9e067ba53ead848ea9a59
--- /dev/null
+++ b/annotator/midas/utils.py
@@ -0,0 +1,189 @@
+"""Utils for monoDepth."""
+import sys
+import re
+import numpy as np
+import cv2
+import torch
+
+
+def read_pfm(path):
+    """Read pfm file.
+
+    Args:
+        path (str): path to file
+
+    Returns:
+        tuple: (data, scale)
+    """
+    with open(path, "rb") as file:
+
+        color = None
+        width = None
+        height = None
+        scale = None
+        endian = None
+
+        header = file.readline().rstrip()
+        if header.decode("ascii") == "PF":
+            color = True
+        elif header.decode("ascii") == "Pf":
+            color = False
+        else:
+            raise Exception("Not a PFM file: " + path)
+
+        dim_match = re.match(r"^(\d+)\s(\d+)\s$", file.readline().decode("ascii"))
+        if dim_match:
+            width, height = list(map(int, dim_match.groups()))
+        else:
+            raise Exception("Malformed PFM header.")
+
+        scale = float(file.readline().decode("ascii").rstrip())
+        if scale < 0:
+            # little-endian
+            endian = "<"
+            scale = -scale
+        else:
+            # big-endian
+            endian = ">"
+
+        data = np.fromfile(file, endian + "f")
+        shape = (height, width, 3) if color else (height, width)
+
+        data = np.reshape(data, shape)
+        data = np.flipud(data)
+
+        return data, scale
+
+
+def write_pfm(path, image, scale=1):
+    """Write pfm file.
+
+    Args:
+        path (str): pathto file
+        image (array): data
+        scale (int, optional): Scale. Defaults to 1.
+    """
+
+    with open(path, "wb") as file:
+        color = None
+
+        if image.dtype.name != "float32":
+            raise Exception("Image dtype must be float32.")
+
+        image = np.flipud(image)
+
+        if len(image.shape) == 3 and image.shape[2] == 3:  # color image
+            color = True
+        elif (
+            len(image.shape) == 2 or len(image.shape) == 3 and image.shape[2] == 1
+        ):  # greyscale
+            color = False
+        else:
+            raise Exception("Image must have H x W x 3, H x W x 1 or H x W dimensions.")
+
+        file.write("PF\n" if color else "Pf\n".encode())
+        file.write("%d %d\n".encode() % (image.shape[1], image.shape[0]))
+
+        endian = image.dtype.byteorder
+
+        if endian == "<" or endian == "=" and sys.byteorder == "little":
+            scale = -scale
+
+        file.write("%f\n".encode() % scale)
+
+        image.tofile(file)
+
+
+def read_image(path):
+    """Read image and output RGB image (0-1).
+
+    Args:
+        path (str): path to file
+
+    Returns:
+        array: RGB image (0-1)
+    """
+    img = cv2.imread(path)
+
+    if img.ndim == 2:
+        img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
+
+    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) / 255.0
+
+    return img
+
+
+def resize_image(img):
+    """Resize image and make it fit for network.
+
+    Args:
+        img (array): image
+
+    Returns:
+        tensor: data ready for network
+    """
+    height_orig = img.shape[0]
+    width_orig = img.shape[1]
+
+    if width_orig > height_orig:
+        scale = width_orig / 384
+    else:
+        scale = height_orig / 384
+
+    height = (np.ceil(height_orig / scale / 32) * 32).astype(int)
+    width = (np.ceil(width_orig / scale / 32) * 32).astype(int)
+
+    img_resized = cv2.resize(img, (width, height), interpolation=cv2.INTER_AREA)
+
+    img_resized = (
+        torch.from_numpy(np.transpose(img_resized, (2, 0, 1))).contiguous().float()
+    )
+    img_resized = img_resized.unsqueeze(0)
+
+    return img_resized
+
+
+def resize_depth(depth, width, height):
+    """Resize depth map and bring to CPU (numpy).
+
+    Args:
+        depth (tensor): depth
+        width (int): image width
+        height (int): image height
+
+    Returns:
+        array: processed depth
+    """
+    depth = torch.squeeze(depth[0, :, :, :]).to("cpu")
+
+    depth_resized = cv2.resize(
+        depth.numpy(), (width, height), interpolation=cv2.INTER_CUBIC
+    )
+
+    return depth_resized
+
+def write_depth(path, depth, bits=1):
+    """Write depth map to pfm and png file.
+
+    Args:
+        path (str): filepath without extension
+        depth (array): depth
+    """
+    write_pfm(path + ".pfm", depth.astype(np.float32))
+
+    depth_min = depth.min()
+    depth_max = depth.max()
+
+    max_val = (2**(8*bits))-1
+
+    if depth_max - depth_min > np.finfo("float").eps:
+        out = max_val * (depth - depth_min) / (depth_max - depth_min)
+    else:
+        out = np.zeros(depth.shape, dtype=depth.type)
+
+    if bits == 1:
+        cv2.imwrite(path + ".png", out.astype("uint8"))
+    elif bits == 2:
+        cv2.imwrite(path + ".png", out.astype("uint16"))
+
+    return
diff --git a/annotator/mlsd/__init__.py b/annotator/mlsd/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..42af28c682e781b30f691f65a475b53c9f3adc8b
--- /dev/null
+++ b/annotator/mlsd/__init__.py
@@ -0,0 +1,39 @@
+import cv2
+import numpy as np
+import torch
+import os
+
+from einops import rearrange
+from .models.mbv2_mlsd_tiny import MobileV2_MLSD_Tiny
+from .models.mbv2_mlsd_large import MobileV2_MLSD_Large
+from .utils import pred_lines
+
+from annotator.util import annotator_ckpts_path
+
+
+remote_model_path = "https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/mlsd_large_512_fp32.pth"
+
+
+class MLSDdetector:
+    def __init__(self):
+        model_path = os.path.join(annotator_ckpts_path, "mlsd_large_512_fp32.pth")
+        if not os.path.exists(model_path):
+            from basicsr.utils.download_util import load_file_from_url
+            load_file_from_url(remote_model_path, model_dir=annotator_ckpts_path)
+        model = MobileV2_MLSD_Large()
+        model.load_state_dict(torch.load(model_path), strict=True)
+        self.model = model.cuda().eval()
+
+    def __call__(self, input_image, thr_v, thr_d):
+        assert input_image.ndim == 3
+        img = input_image
+        img_output = np.zeros_like(img)
+        try:
+            with torch.no_grad():
+                lines = pred_lines(img, self.model, [img.shape[0], img.shape[1]], thr_v, thr_d)
+                for line in lines:
+                    x_start, y_start, x_end, y_end = [int(val) for val in line]
+                    cv2.line(img_output, (x_start, y_start), (x_end, y_end), [255, 255, 255], 1)
+        except Exception as e:
+            pass
+        return img_output[:, :, 0]
diff --git a/annotator/mlsd/models/mbv2_mlsd_large.py b/annotator/mlsd/models/mbv2_mlsd_large.py
new file mode 100644
index 0000000000000000000000000000000000000000..5b9799e7573ca41549b3c3b13ac47b906b369603
--- /dev/null
+++ b/annotator/mlsd/models/mbv2_mlsd_large.py
@@ -0,0 +1,292 @@
+import os
+import sys
+import torch
+import torch.nn as nn
+import torch.utils.model_zoo as model_zoo
+from  torch.nn import  functional as F
+
+
+class BlockTypeA(nn.Module):
+    def __init__(self, in_c1, in_c2, out_c1, out_c2, upscale = True):
+        super(BlockTypeA, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_c2, out_c2, kernel_size=1),
+            nn.BatchNorm2d(out_c2),
+            nn.ReLU(inplace=True)
+        )
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_c1, out_c1, kernel_size=1),
+            nn.BatchNorm2d(out_c1),
+            nn.ReLU(inplace=True)
+        )
+        self.upscale = upscale
+
+    def forward(self, a, b):
+        b = self.conv1(b)
+        a = self.conv2(a)
+        if self.upscale:
+             b = F.interpolate(b, scale_factor=2.0, mode='bilinear', align_corners=True)
+        return torch.cat((a, b), dim=1)
+
+
+class BlockTypeB(nn.Module):
+    def __init__(self, in_c, out_c):
+        super(BlockTypeB, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_c, in_c,  kernel_size=3, padding=1),
+            nn.BatchNorm2d(in_c),
+            nn.ReLU()
+        )
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_c, out_c, kernel_size=3, padding=1),
+            nn.BatchNorm2d(out_c),
+            nn.ReLU()
+        )
+
+    def forward(self, x):
+        x = self.conv1(x) + x
+        x = self.conv2(x)
+        return x
+
+class BlockTypeC(nn.Module):
+    def __init__(self, in_c, out_c):
+        super(BlockTypeC, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_c, in_c,  kernel_size=3, padding=5, dilation=5),
+            nn.BatchNorm2d(in_c),
+            nn.ReLU()
+        )
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_c, in_c,  kernel_size=3, padding=1),
+            nn.BatchNorm2d(in_c),
+            nn.ReLU()
+        )
+        self.conv3 = nn.Conv2d(in_c, out_c, kernel_size=1)
+
+    def forward(self, x):
+        x = self.conv1(x)
+        x = self.conv2(x)
+        x = self.conv3(x)
+        return x
+
+def _make_divisible(v, divisor, min_value=None):
+    """
+    This function is taken from the original tf repo.
+    It ensures that all layers have a channel number that is divisible by 8
+    It can be seen here:
+    https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet/mobilenet.py
+    :param v:
+    :param divisor:
+    :param min_value:
+    :return:
+    """
+    if min_value is None:
+        min_value = divisor
+    new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
+    # Make sure that round down does not go down by more than 10%.
+    if new_v < 0.9 * v:
+        new_v += divisor
+    return new_v
+
+
+class ConvBNReLU(nn.Sequential):
+    def __init__(self, in_planes, out_planes, kernel_size=3, stride=1, groups=1):
+        self.channel_pad = out_planes - in_planes
+        self.stride = stride
+        #padding = (kernel_size - 1) // 2
+
+        # TFLite uses slightly different padding than PyTorch
+        if stride == 2:
+            padding = 0
+        else:
+            padding = (kernel_size - 1) // 2
+
+        super(ConvBNReLU, self).__init__(
+            nn.Conv2d(in_planes, out_planes, kernel_size, stride, padding, groups=groups, bias=False),
+            nn.BatchNorm2d(out_planes),
+            nn.ReLU6(inplace=True)
+        )
+        self.max_pool = nn.MaxPool2d(kernel_size=stride, stride=stride)
+
+
+    def forward(self, x):
+        # TFLite uses  different padding
+        if self.stride == 2:
+            x = F.pad(x, (0, 1, 0, 1), "constant", 0)
+            #print(x.shape)
+
+        for module in self:
+            if not isinstance(module, nn.MaxPool2d):
+                x = module(x)
+        return x
+
+
+class InvertedResidual(nn.Module):
+    def __init__(self, inp, oup, stride, expand_ratio):
+        super(InvertedResidual, self).__init__()
+        self.stride = stride
+        assert stride in [1, 2]
+
+        hidden_dim = int(round(inp * expand_ratio))
+        self.use_res_connect = self.stride == 1 and inp == oup
+
+        layers = []
+        if expand_ratio != 1:
+            # pw
+            layers.append(ConvBNReLU(inp, hidden_dim, kernel_size=1))
+        layers.extend([
+            # dw
+            ConvBNReLU(hidden_dim, hidden_dim, stride=stride, groups=hidden_dim),
+            # pw-linear
+            nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
+            nn.BatchNorm2d(oup),
+        ])
+        self.conv = nn.Sequential(*layers)
+
+    def forward(self, x):
+        if self.use_res_connect:
+            return x + self.conv(x)
+        else:
+            return self.conv(x)
+
+
+class MobileNetV2(nn.Module):
+    def __init__(self, pretrained=True):
+        """
+        MobileNet V2 main class
+        Args:
+            num_classes (int): Number of classes
+            width_mult (float): Width multiplier - adjusts number of channels in each layer by this amount
+            inverted_residual_setting: Network structure
+            round_nearest (int): Round the number of channels in each layer to be a multiple of this number
+            Set to 1 to turn off rounding
+            block: Module specifying inverted residual building block for mobilenet
+        """
+        super(MobileNetV2, self).__init__()
+
+        block = InvertedResidual
+        input_channel = 32
+        last_channel = 1280
+        width_mult = 1.0
+        round_nearest = 8
+
+        inverted_residual_setting = [
+            # t, c, n, s
+            [1, 16, 1, 1],
+            [6, 24, 2, 2],
+            [6, 32, 3, 2],
+            [6, 64, 4, 2],
+            [6, 96, 3, 1],
+            #[6, 160, 3, 2],
+            #[6, 320, 1, 1],
+        ]
+
+        # only check the first element, assuming user knows t,c,n,s are required
+        if len(inverted_residual_setting) == 0 or len(inverted_residual_setting[0]) != 4:
+            raise ValueError("inverted_residual_setting should be non-empty "
+                             "or a 4-element list, got {}".format(inverted_residual_setting))
+
+        # building first layer
+        input_channel = _make_divisible(input_channel * width_mult, round_nearest)
+        self.last_channel = _make_divisible(last_channel * max(1.0, width_mult), round_nearest)
+        features = [ConvBNReLU(4, input_channel, stride=2)]
+        # building inverted residual blocks
+        for t, c, n, s in inverted_residual_setting:
+            output_channel = _make_divisible(c * width_mult, round_nearest)
+            for i in range(n):
+                stride = s if i == 0 else 1
+                features.append(block(input_channel, output_channel, stride, expand_ratio=t))
+                input_channel = output_channel
+
+        self.features = nn.Sequential(*features)
+        self.fpn_selected = [1, 3, 6, 10, 13]
+        # weight initialization
+        for m in self.modules():
+            if isinstance(m, nn.Conv2d):
+                nn.init.kaiming_normal_(m.weight, mode='fan_out')
+                if m.bias is not None:
+                    nn.init.zeros_(m.bias)
+            elif isinstance(m, nn.BatchNorm2d):
+                nn.init.ones_(m.weight)
+                nn.init.zeros_(m.bias)
+            elif isinstance(m, nn.Linear):
+                nn.init.normal_(m.weight, 0, 0.01)
+                nn.init.zeros_(m.bias)
+        if pretrained:
+           self._load_pretrained_model()
+
+    def _forward_impl(self, x):
+        # This exists since TorchScript doesn't support inheritance, so the superclass method
+        # (this one) needs to have a name other than `forward` that can be accessed in a subclass
+        fpn_features = []
+        for i, f in enumerate(self.features):
+            if i > self.fpn_selected[-1]:
+                break
+            x = f(x)
+            if i in self.fpn_selected:
+                fpn_features.append(x)
+
+        c1, c2, c3, c4, c5 = fpn_features
+        return c1, c2, c3, c4, c5
+
+
+    def forward(self, x):
+        return self._forward_impl(x)
+
+    def _load_pretrained_model(self):
+        pretrain_dict = model_zoo.load_url('https://download.pytorch.org/models/mobilenet_v2-b0353104.pth')
+        model_dict = {}
+        state_dict = self.state_dict()
+        for k, v in pretrain_dict.items():
+            if k in state_dict:
+                model_dict[k] = v
+        state_dict.update(model_dict)
+        self.load_state_dict(state_dict)
+
+
+class MobileV2_MLSD_Large(nn.Module):
+    def __init__(self):
+        super(MobileV2_MLSD_Large, self).__init__()
+
+        self.backbone = MobileNetV2(pretrained=False)
+        ## A, B
+        self.block15 = BlockTypeA(in_c1= 64, in_c2= 96,
+                                  out_c1= 64, out_c2=64,
+                                  upscale=False)
+        self.block16 = BlockTypeB(128, 64)
+
+        ## A, B
+        self.block17 = BlockTypeA(in_c1 = 32,  in_c2 = 64,
+                                  out_c1= 64,  out_c2= 64)
+        self.block18 = BlockTypeB(128, 64)
+
+        ## A, B
+        self.block19 = BlockTypeA(in_c1=24, in_c2=64,
+                                  out_c1=64, out_c2=64)
+        self.block20 = BlockTypeB(128, 64)
+
+        ## A, B, C
+        self.block21 = BlockTypeA(in_c1=16, in_c2=64,
+                                  out_c1=64, out_c2=64)
+        self.block22 = BlockTypeB(128, 64)
+
+        self.block23 = BlockTypeC(64, 16)
+
+    def forward(self, x):
+        c1, c2, c3, c4, c5 = self.backbone(x)
+
+        x = self.block15(c4, c5)
+        x = self.block16(x)
+
+        x = self.block17(c3, x)
+        x = self.block18(x)
+
+        x = self.block19(c2, x)
+        x = self.block20(x)
+
+        x = self.block21(c1, x)
+        x = self.block22(x)
+        x = self.block23(x)
+        x = x[:, 7:, :, :]
+
+        return x
\ No newline at end of file
diff --git a/annotator/mlsd/models/mbv2_mlsd_tiny.py b/annotator/mlsd/models/mbv2_mlsd_tiny.py
new file mode 100644
index 0000000000000000000000000000000000000000..e3ed633f2cc23ea1829a627fdb879ab39f641f83
--- /dev/null
+++ b/annotator/mlsd/models/mbv2_mlsd_tiny.py
@@ -0,0 +1,275 @@
+import os
+import sys
+import torch
+import torch.nn as nn
+import torch.utils.model_zoo as model_zoo
+from  torch.nn import  functional as F
+
+
+class BlockTypeA(nn.Module):
+    def __init__(self, in_c1, in_c2, out_c1, out_c2, upscale = True):
+        super(BlockTypeA, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_c2, out_c2, kernel_size=1),
+            nn.BatchNorm2d(out_c2),
+            nn.ReLU(inplace=True)
+        )
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_c1, out_c1, kernel_size=1),
+            nn.BatchNorm2d(out_c1),
+            nn.ReLU(inplace=True)
+        )
+        self.upscale = upscale
+
+    def forward(self, a, b):
+        b = self.conv1(b)
+        a = self.conv2(a)
+        b = F.interpolate(b, scale_factor=2.0, mode='bilinear', align_corners=True)
+        return torch.cat((a, b), dim=1)
+
+
+class BlockTypeB(nn.Module):
+    def __init__(self, in_c, out_c):
+        super(BlockTypeB, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_c, in_c,  kernel_size=3, padding=1),
+            nn.BatchNorm2d(in_c),
+            nn.ReLU()
+        )
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_c, out_c, kernel_size=3, padding=1),
+            nn.BatchNorm2d(out_c),
+            nn.ReLU()
+        )
+
+    def forward(self, x):
+        x = self.conv1(x) + x
+        x = self.conv2(x)
+        return x
+
+class BlockTypeC(nn.Module):
+    def __init__(self, in_c, out_c):
+        super(BlockTypeC, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_c, in_c,  kernel_size=3, padding=5, dilation=5),
+            nn.BatchNorm2d(in_c),
+            nn.ReLU()
+        )
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_c, in_c,  kernel_size=3, padding=1),
+            nn.BatchNorm2d(in_c),
+            nn.ReLU()
+        )
+        self.conv3 = nn.Conv2d(in_c, out_c, kernel_size=1)
+
+    def forward(self, x):
+        x = self.conv1(x)
+        x = self.conv2(x)
+        x = self.conv3(x)
+        return x
+
+def _make_divisible(v, divisor, min_value=None):
+    """
+    This function is taken from the original tf repo.
+    It ensures that all layers have a channel number that is divisible by 8
+    It can be seen here:
+    https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet/mobilenet.py
+    :param v:
+    :param divisor:
+    :param min_value:
+    :return:
+    """
+    if min_value is None:
+        min_value = divisor
+    new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
+    # Make sure that round down does not go down by more than 10%.
+    if new_v < 0.9 * v:
+        new_v += divisor
+    return new_v
+
+
+class ConvBNReLU(nn.Sequential):
+    def __init__(self, in_planes, out_planes, kernel_size=3, stride=1, groups=1):
+        self.channel_pad = out_planes - in_planes
+        self.stride = stride
+        #padding = (kernel_size - 1) // 2
+
+        # TFLite uses slightly different padding than PyTorch
+        if stride == 2:
+            padding = 0
+        else:
+            padding = (kernel_size - 1) // 2
+
+        super(ConvBNReLU, self).__init__(
+            nn.Conv2d(in_planes, out_planes, kernel_size, stride, padding, groups=groups, bias=False),
+            nn.BatchNorm2d(out_planes),
+            nn.ReLU6(inplace=True)
+        )
+        self.max_pool = nn.MaxPool2d(kernel_size=stride, stride=stride)
+
+
+    def forward(self, x):
+        # TFLite uses  different padding
+        if self.stride == 2:
+            x = F.pad(x, (0, 1, 0, 1), "constant", 0)
+            #print(x.shape)
+
+        for module in self:
+            if not isinstance(module, nn.MaxPool2d):
+                x = module(x)
+        return x
+
+
+class InvertedResidual(nn.Module):
+    def __init__(self, inp, oup, stride, expand_ratio):
+        super(InvertedResidual, self).__init__()
+        self.stride = stride
+        assert stride in [1, 2]
+
+        hidden_dim = int(round(inp * expand_ratio))
+        self.use_res_connect = self.stride == 1 and inp == oup
+
+        layers = []
+        if expand_ratio != 1:
+            # pw
+            layers.append(ConvBNReLU(inp, hidden_dim, kernel_size=1))
+        layers.extend([
+            # dw
+            ConvBNReLU(hidden_dim, hidden_dim, stride=stride, groups=hidden_dim),
+            # pw-linear
+            nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
+            nn.BatchNorm2d(oup),
+        ])
+        self.conv = nn.Sequential(*layers)
+
+    def forward(self, x):
+        if self.use_res_connect:
+            return x + self.conv(x)
+        else:
+            return self.conv(x)
+
+
+class MobileNetV2(nn.Module):
+    def __init__(self, pretrained=True):
+        """
+        MobileNet V2 main class
+        Args:
+            num_classes (int): Number of classes
+            width_mult (float): Width multiplier - adjusts number of channels in each layer by this amount
+            inverted_residual_setting: Network structure
+            round_nearest (int): Round the number of channels in each layer to be a multiple of this number
+            Set to 1 to turn off rounding
+            block: Module specifying inverted residual building block for mobilenet
+        """
+        super(MobileNetV2, self).__init__()
+
+        block = InvertedResidual
+        input_channel = 32
+        last_channel = 1280
+        width_mult = 1.0
+        round_nearest = 8
+
+        inverted_residual_setting = [
+            # t, c, n, s
+            [1, 16, 1, 1],
+            [6, 24, 2, 2],
+            [6, 32, 3, 2],
+            [6, 64, 4, 2],
+            #[6, 96, 3, 1],
+            #[6, 160, 3, 2],
+            #[6, 320, 1, 1],
+        ]
+
+        # only check the first element, assuming user knows t,c,n,s are required
+        if len(inverted_residual_setting) == 0 or len(inverted_residual_setting[0]) != 4:
+            raise ValueError("inverted_residual_setting should be non-empty "
+                             "or a 4-element list, got {}".format(inverted_residual_setting))
+
+        # building first layer
+        input_channel = _make_divisible(input_channel * width_mult, round_nearest)
+        self.last_channel = _make_divisible(last_channel * max(1.0, width_mult), round_nearest)
+        features = [ConvBNReLU(4, input_channel, stride=2)]
+        # building inverted residual blocks
+        for t, c, n, s in inverted_residual_setting:
+            output_channel = _make_divisible(c * width_mult, round_nearest)
+            for i in range(n):
+                stride = s if i == 0 else 1
+                features.append(block(input_channel, output_channel, stride, expand_ratio=t))
+                input_channel = output_channel
+        self.features = nn.Sequential(*features)
+
+        self.fpn_selected = [3, 6, 10]
+        # weight initialization
+        for m in self.modules():
+            if isinstance(m, nn.Conv2d):
+                nn.init.kaiming_normal_(m.weight, mode='fan_out')
+                if m.bias is not None:
+                    nn.init.zeros_(m.bias)
+            elif isinstance(m, nn.BatchNorm2d):
+                nn.init.ones_(m.weight)
+                nn.init.zeros_(m.bias)
+            elif isinstance(m, nn.Linear):
+                nn.init.normal_(m.weight, 0, 0.01)
+                nn.init.zeros_(m.bias)
+
+        #if pretrained:
+        #    self._load_pretrained_model()
+
+    def _forward_impl(self, x):
+        # This exists since TorchScript doesn't support inheritance, so the superclass method
+        # (this one) needs to have a name other than `forward` that can be accessed in a subclass
+        fpn_features = []
+        for i, f in enumerate(self.features):
+            if i > self.fpn_selected[-1]:
+                break
+            x = f(x)
+            if i in self.fpn_selected:
+                fpn_features.append(x)
+
+        c2, c3, c4 = fpn_features
+        return c2, c3, c4
+
+
+    def forward(self, x):
+        return self._forward_impl(x)
+
+    def _load_pretrained_model(self):
+        pretrain_dict = model_zoo.load_url('https://download.pytorch.org/models/mobilenet_v2-b0353104.pth')
+        model_dict = {}
+        state_dict = self.state_dict()
+        for k, v in pretrain_dict.items():
+            if k in state_dict:
+                model_dict[k] = v
+        state_dict.update(model_dict)
+        self.load_state_dict(state_dict)
+
+
+class MobileV2_MLSD_Tiny(nn.Module):
+    def __init__(self):
+        super(MobileV2_MLSD_Tiny, self).__init__()
+
+        self.backbone = MobileNetV2(pretrained=True)
+
+        self.block12 = BlockTypeA(in_c1= 32, in_c2= 64,
+                                  out_c1= 64, out_c2=64)
+        self.block13 = BlockTypeB(128, 64)
+
+        self.block14 = BlockTypeA(in_c1 = 24,  in_c2 = 64,
+                                  out_c1= 32,  out_c2= 32)
+        self.block15 = BlockTypeB(64, 64)
+
+        self.block16 = BlockTypeC(64, 16)
+
+    def forward(self, x):
+        c2, c3, c4 = self.backbone(x)
+
+        x = self.block12(c3, c4)
+        x = self.block13(x)
+        x = self.block14(c2, x)
+        x = self.block15(x)
+        x = self.block16(x)
+        x = x[:, 7:, :, :]
+        #print(x.shape)
+        x = F.interpolate(x, scale_factor=2.0, mode='bilinear', align_corners=True)
+
+        return x
\ No newline at end of file
diff --git a/annotator/mlsd/utils.py b/annotator/mlsd/utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..ae3cf9420a33a4abae27c48ac4b90938c7d63cc3
--- /dev/null
+++ b/annotator/mlsd/utils.py
@@ -0,0 +1,580 @@
+'''
+modified by  lihaoweicv
+pytorch version
+'''
+
+'''
+M-LSD
+Copyright 2021-present NAVER Corp.
+Apache License v2.0
+'''
+
+import os
+import numpy as np
+import cv2
+import torch
+from  torch.nn import  functional as F
+
+
+def deccode_output_score_and_ptss(tpMap, topk_n = 200, ksize = 5):
+    '''
+    tpMap:
+    center: tpMap[1, 0, :, :]
+    displacement: tpMap[1, 1:5, :, :]
+    '''
+    b, c, h, w = tpMap.shape
+    assert  b==1, 'only support bsize==1'
+    displacement = tpMap[:, 1:5, :, :][0]
+    center = tpMap[:, 0, :, :]
+    heat = torch.sigmoid(center)
+    hmax = F.max_pool2d( heat, (ksize, ksize), stride=1, padding=(ksize-1)//2)
+    keep = (hmax == heat).float()
+    heat = heat * keep
+    heat = heat.reshape(-1, )
+
+    scores, indices = torch.topk(heat, topk_n, dim=-1, largest=True)
+    yy = torch.floor_divide(indices, w).unsqueeze(-1)
+    xx = torch.fmod(indices, w).unsqueeze(-1)
+    ptss = torch.cat((yy, xx),dim=-1)
+
+    ptss   = ptss.detach().cpu().numpy()
+    scores = scores.detach().cpu().numpy()
+    displacement = displacement.detach().cpu().numpy()
+    displacement = displacement.transpose((1,2,0))
+    return  ptss, scores, displacement
+
+
+def pred_lines(image, model,
+               input_shape=[512, 512],
+               score_thr=0.10,
+               dist_thr=20.0):
+    h, w, _ = image.shape
+    h_ratio, w_ratio = [h / input_shape[0], w / input_shape[1]]
+
+    resized_image = np.concatenate([cv2.resize(image, (input_shape[1], input_shape[0]), interpolation=cv2.INTER_AREA),
+                                    np.ones([input_shape[0], input_shape[1], 1])], axis=-1)
+
+    resized_image = resized_image.transpose((2,0,1))
+    batch_image = np.expand_dims(resized_image, axis=0).astype('float32')
+    batch_image = (batch_image / 127.5) - 1.0
+
+    batch_image = torch.from_numpy(batch_image).float().cuda()
+    outputs = model(batch_image)
+    pts, pts_score, vmap = deccode_output_score_and_ptss(outputs, 200, 3)
+    start = vmap[:, :, :2]
+    end = vmap[:, :, 2:]
+    dist_map = np.sqrt(np.sum((start - end) ** 2, axis=-1))
+
+    segments_list = []
+    for center, score in zip(pts, pts_score):
+        y, x = center
+        distance = dist_map[y, x]
+        if score > score_thr and distance > dist_thr:
+            disp_x_start, disp_y_start, disp_x_end, disp_y_end = vmap[y, x, :]
+            x_start = x + disp_x_start
+            y_start = y + disp_y_start
+            x_end = x + disp_x_end
+            y_end = y + disp_y_end
+            segments_list.append([x_start, y_start, x_end, y_end])
+
+    lines = 2 * np.array(segments_list)  # 256 > 512
+    lines[:, 0] = lines[:, 0] * w_ratio
+    lines[:, 1] = lines[:, 1] * h_ratio
+    lines[:, 2] = lines[:, 2] * w_ratio
+    lines[:, 3] = lines[:, 3] * h_ratio
+
+    return lines
+
+
+def pred_squares(image,
+                 model,
+                 input_shape=[512, 512],
+                 params={'score': 0.06,
+                         'outside_ratio': 0.28,
+                         'inside_ratio': 0.45,
+                         'w_overlap': 0.0,
+                         'w_degree': 1.95,
+                         'w_length': 0.0,
+                         'w_area': 1.86,
+                         'w_center': 0.14}):
+    '''
+    shape = [height, width]
+    '''
+    h, w, _ = image.shape
+    original_shape = [h, w]
+
+    resized_image = np.concatenate([cv2.resize(image, (input_shape[0], input_shape[1]), interpolation=cv2.INTER_AREA),
+                                    np.ones([input_shape[0], input_shape[1], 1])], axis=-1)
+    resized_image = resized_image.transpose((2, 0, 1))
+    batch_image = np.expand_dims(resized_image, axis=0).astype('float32')
+    batch_image = (batch_image / 127.5) - 1.0
+
+    batch_image = torch.from_numpy(batch_image).float().cuda()
+    outputs = model(batch_image)
+
+    pts, pts_score, vmap = deccode_output_score_and_ptss(outputs, 200, 3)
+    start = vmap[:, :, :2]  # (x, y)
+    end = vmap[:, :, 2:]  # (x, y)
+    dist_map = np.sqrt(np.sum((start - end) ** 2, axis=-1))
+
+    junc_list = []
+    segments_list = []
+    for junc, score in zip(pts, pts_score):
+        y, x = junc
+        distance = dist_map[y, x]
+        if score > params['score'] and distance > 20.0:
+            junc_list.append([x, y])
+            disp_x_start, disp_y_start, disp_x_end, disp_y_end = vmap[y, x, :]
+            d_arrow = 1.0
+            x_start = x + d_arrow * disp_x_start
+            y_start = y + d_arrow * disp_y_start
+            x_end = x + d_arrow * disp_x_end
+            y_end = y + d_arrow * disp_y_end
+            segments_list.append([x_start, y_start, x_end, y_end])
+
+    segments = np.array(segments_list)
+
+    ####### post processing for squares
+    # 1. get unique lines
+    point = np.array([[0, 0]])
+    point = point[0]
+    start = segments[:, :2]
+    end = segments[:, 2:]
+    diff = start - end
+    a = diff[:, 1]
+    b = -diff[:, 0]
+    c = a * start[:, 0] + b * start[:, 1]
+
+    d = np.abs(a * point[0] + b * point[1] - c) / np.sqrt(a ** 2 + b ** 2 + 1e-10)
+    theta = np.arctan2(diff[:, 0], diff[:, 1]) * 180 / np.pi
+    theta[theta < 0.0] += 180
+    hough = np.concatenate([d[:, None], theta[:, None]], axis=-1)
+
+    d_quant = 1
+    theta_quant = 2
+    hough[:, 0] //= d_quant
+    hough[:, 1] //= theta_quant
+    _, indices, counts = np.unique(hough, axis=0, return_index=True, return_counts=True)
+
+    acc_map = np.zeros([512 // d_quant + 1, 360 // theta_quant + 1], dtype='float32')
+    idx_map = np.zeros([512 // d_quant + 1, 360 // theta_quant + 1], dtype='int32') - 1
+    yx_indices = hough[indices, :].astype('int32')
+    acc_map[yx_indices[:, 0], yx_indices[:, 1]] = counts
+    idx_map[yx_indices[:, 0], yx_indices[:, 1]] = indices
+
+    acc_map_np = acc_map
+    # acc_map = acc_map[None, :, :, None]
+    #
+    # ### fast suppression using tensorflow op
+    # acc_map = tf.constant(acc_map, dtype=tf.float32)
+    # max_acc_map = tf.keras.layers.MaxPool2D(pool_size=(5, 5), strides=1, padding='same')(acc_map)
+    # acc_map = acc_map * tf.cast(tf.math.equal(acc_map, max_acc_map), tf.float32)
+    # flatten_acc_map = tf.reshape(acc_map, [1, -1])
+    # topk_values, topk_indices = tf.math.top_k(flatten_acc_map, k=len(pts))
+    # _, h, w, _ = acc_map.shape
+    # y = tf.expand_dims(topk_indices // w, axis=-1)
+    # x = tf.expand_dims(topk_indices % w, axis=-1)
+    # yx = tf.concat([y, x], axis=-1)
+
+    ### fast suppression using pytorch op
+    acc_map = torch.from_numpy(acc_map_np).unsqueeze(0).unsqueeze(0)
+    _,_, h, w = acc_map.shape
+    max_acc_map = F.max_pool2d(acc_map,kernel_size=5, stride=1, padding=2)
+    acc_map = acc_map * ( (acc_map == max_acc_map).float() )
+    flatten_acc_map = acc_map.reshape([-1, ])
+
+    scores, indices = torch.topk(flatten_acc_map, len(pts), dim=-1, largest=True)
+    yy = torch.div(indices, w, rounding_mode='floor').unsqueeze(-1)
+    xx = torch.fmod(indices, w).unsqueeze(-1)
+    yx = torch.cat((yy, xx), dim=-1)
+
+    yx = yx.detach().cpu().numpy()
+
+    topk_values = scores.detach().cpu().numpy()
+    indices = idx_map[yx[:, 0], yx[:, 1]]
+    basis = 5 // 2
+
+    merged_segments = []
+    for yx_pt, max_indice, value in zip(yx, indices, topk_values):
+        y, x = yx_pt
+        if max_indice == -1 or value == 0:
+            continue
+        segment_list = []
+        for y_offset in range(-basis, basis + 1):
+            for x_offset in range(-basis, basis + 1):
+                indice = idx_map[y + y_offset, x + x_offset]
+                cnt = int(acc_map_np[y + y_offset, x + x_offset])
+                if indice != -1:
+                    segment_list.append(segments[indice])
+                if cnt > 1:
+                    check_cnt = 1
+                    current_hough = hough[indice]
+                    for new_indice, new_hough in enumerate(hough):
+                        if (current_hough == new_hough).all() and indice != new_indice:
+                            segment_list.append(segments[new_indice])
+                            check_cnt += 1
+                        if check_cnt == cnt:
+                            break
+        group_segments = np.array(segment_list).reshape([-1, 2])
+        sorted_group_segments = np.sort(group_segments, axis=0)
+        x_min, y_min = sorted_group_segments[0, :]
+        x_max, y_max = sorted_group_segments[-1, :]
+
+        deg = theta[max_indice]
+        if deg >= 90:
+            merged_segments.append([x_min, y_max, x_max, y_min])
+        else:
+            merged_segments.append([x_min, y_min, x_max, y_max])
+
+    # 2. get intersections
+    new_segments = np.array(merged_segments)  # (x1, y1, x2, y2)
+    start = new_segments[:, :2]  # (x1, y1)
+    end = new_segments[:, 2:]  # (x2, y2)
+    new_centers = (start + end) / 2.0
+    diff = start - end
+    dist_segments = np.sqrt(np.sum(diff ** 2, axis=-1))
+
+    # ax + by = c
+    a = diff[:, 1]
+    b = -diff[:, 0]
+    c = a * start[:, 0] + b * start[:, 1]
+    pre_det = a[:, None] * b[None, :]
+    det = pre_det - np.transpose(pre_det)
+
+    pre_inter_y = a[:, None] * c[None, :]
+    inter_y = (pre_inter_y - np.transpose(pre_inter_y)) / (det + 1e-10)
+    pre_inter_x = c[:, None] * b[None, :]
+    inter_x = (pre_inter_x - np.transpose(pre_inter_x)) / (det + 1e-10)
+    inter_pts = np.concatenate([inter_x[:, :, None], inter_y[:, :, None]], axis=-1).astype('int32')
+
+    # 3. get corner information
+    # 3.1 get distance
+    '''
+    dist_segments:
+        | dist(0), dist(1), dist(2), ...|
+    dist_inter_to_segment1:
+        | dist(inter,0), dist(inter,0), dist(inter,0), ... |
+        | dist(inter,1), dist(inter,1), dist(inter,1), ... |
+        ...
+    dist_inter_to_semgnet2:
+        | dist(inter,0), dist(inter,1), dist(inter,2), ... |
+        | dist(inter,0), dist(inter,1), dist(inter,2), ... |
+        ...
+    '''
+
+    dist_inter_to_segment1_start = np.sqrt(
+        np.sum(((inter_pts - start[:, None, :]) ** 2), axis=-1, keepdims=True))  # [n_batch, n_batch, 1]
+    dist_inter_to_segment1_end = np.sqrt(
+        np.sum(((inter_pts - end[:, None, :]) ** 2), axis=-1, keepdims=True))  # [n_batch, n_batch, 1]
+    dist_inter_to_segment2_start = np.sqrt(
+        np.sum(((inter_pts - start[None, :, :]) ** 2), axis=-1, keepdims=True))  # [n_batch, n_batch, 1]
+    dist_inter_to_segment2_end = np.sqrt(
+        np.sum(((inter_pts - end[None, :, :]) ** 2), axis=-1, keepdims=True))  # [n_batch, n_batch, 1]
+
+    # sort ascending
+    dist_inter_to_segment1 = np.sort(
+        np.concatenate([dist_inter_to_segment1_start, dist_inter_to_segment1_end], axis=-1),
+        axis=-1)  # [n_batch, n_batch, 2]
+    dist_inter_to_segment2 = np.sort(
+        np.concatenate([dist_inter_to_segment2_start, dist_inter_to_segment2_end], axis=-1),
+        axis=-1)  # [n_batch, n_batch, 2]
+
+    # 3.2 get degree
+    inter_to_start = new_centers[:, None, :] - inter_pts
+    deg_inter_to_start = np.arctan2(inter_to_start[:, :, 1], inter_to_start[:, :, 0]) * 180 / np.pi
+    deg_inter_to_start[deg_inter_to_start < 0.0] += 360
+    inter_to_end = new_centers[None, :, :] - inter_pts
+    deg_inter_to_end = np.arctan2(inter_to_end[:, :, 1], inter_to_end[:, :, 0]) * 180 / np.pi
+    deg_inter_to_end[deg_inter_to_end < 0.0] += 360
+
+    '''
+    B -- G
+    |    |
+    C -- R
+    B : blue / G: green / C: cyan / R: red
+
+    0 -- 1
+    |    |
+    3 -- 2
+    '''
+    # rename variables
+    deg1_map, deg2_map = deg_inter_to_start, deg_inter_to_end
+    # sort deg ascending
+    deg_sort = np.sort(np.concatenate([deg1_map[:, :, None], deg2_map[:, :, None]], axis=-1), axis=-1)
+
+    deg_diff_map = np.abs(deg1_map - deg2_map)
+    # we only consider the smallest degree of intersect
+    deg_diff_map[deg_diff_map > 180] = 360 - deg_diff_map[deg_diff_map > 180]
+
+    # define available degree range
+    deg_range = [60, 120]
+
+    corner_dict = {corner_info: [] for corner_info in range(4)}
+    inter_points = []
+    for i in range(inter_pts.shape[0]):
+        for j in range(i + 1, inter_pts.shape[1]):
+            # i, j > line index, always i < j
+            x, y = inter_pts[i, j, :]
+            deg1, deg2 = deg_sort[i, j, :]
+            deg_diff = deg_diff_map[i, j]
+
+            check_degree = deg_diff > deg_range[0] and deg_diff < deg_range[1]
+
+            outside_ratio = params['outside_ratio']  # over ratio >>> drop it!
+            inside_ratio = params['inside_ratio']  # over ratio >>> drop it!
+            check_distance = ((dist_inter_to_segment1[i, j, 1] >= dist_segments[i] and \
+                               dist_inter_to_segment1[i, j, 0] <= dist_segments[i] * outside_ratio) or \
+                              (dist_inter_to_segment1[i, j, 1] <= dist_segments[i] and \
+                               dist_inter_to_segment1[i, j, 0] <= dist_segments[i] * inside_ratio)) and \
+                             ((dist_inter_to_segment2[i, j, 1] >= dist_segments[j] and \
+                               dist_inter_to_segment2[i, j, 0] <= dist_segments[j] * outside_ratio) or \
+                              (dist_inter_to_segment2[i, j, 1] <= dist_segments[j] and \
+                               dist_inter_to_segment2[i, j, 0] <= dist_segments[j] * inside_ratio))
+
+            if check_degree and check_distance:
+                corner_info = None
+
+                if (deg1 >= 0 and deg1 <= 45 and deg2 >= 45 and deg2 <= 120) or \
+                        (deg2 >= 315 and deg1 >= 45 and deg1 <= 120):
+                    corner_info, color_info = 0, 'blue'
+                elif (deg1 >= 45 and deg1 <= 125 and deg2 >= 125 and deg2 <= 225):
+                    corner_info, color_info = 1, 'green'
+                elif (deg1 >= 125 and deg1 <= 225 and deg2 >= 225 and deg2 <= 315):
+                    corner_info, color_info = 2, 'black'
+                elif (deg1 >= 0 and deg1 <= 45 and deg2 >= 225 and deg2 <= 315) or \
+                        (deg2 >= 315 and deg1 >= 225 and deg1 <= 315):
+                    corner_info, color_info = 3, 'cyan'
+                else:
+                    corner_info, color_info = 4, 'red'  # we don't use it
+                    continue
+
+                corner_dict[corner_info].append([x, y, i, j])
+                inter_points.append([x, y])
+
+    square_list = []
+    connect_list = []
+    segments_list = []
+    for corner0 in corner_dict[0]:
+        for corner1 in corner_dict[1]:
+            connect01 = False
+            for corner0_line in corner0[2:]:
+                if corner0_line in corner1[2:]:
+                    connect01 = True
+                    break
+            if connect01:
+                for corner2 in corner_dict[2]:
+                    connect12 = False
+                    for corner1_line in corner1[2:]:
+                        if corner1_line in corner2[2:]:
+                            connect12 = True
+                            break
+                    if connect12:
+                        for corner3 in corner_dict[3]:
+                            connect23 = False
+                            for corner2_line in corner2[2:]:
+                                if corner2_line in corner3[2:]:
+                                    connect23 = True
+                                    break
+                            if connect23:
+                                for corner3_line in corner3[2:]:
+                                    if corner3_line in corner0[2:]:
+                                        # SQUARE!!!
+                                        '''
+                                        0 -- 1
+                                        |    |
+                                        3 -- 2
+                                        square_list:
+                                            order: 0 > 1 > 2 > 3
+                                            | x0, y0, x1, y1, x2, y2, x3, y3 |
+                                            | x0, y0, x1, y1, x2, y2, x3, y3 |
+                                            ...
+                                        connect_list:
+                                            order: 01 > 12 > 23 > 30
+                                            | line_idx01, line_idx12, line_idx23, line_idx30 |
+                                            | line_idx01, line_idx12, line_idx23, line_idx30 |
+                                            ...
+                                        segments_list:
+                                            order: 0 > 1 > 2 > 3
+                                            | line_idx0_i, line_idx0_j, line_idx1_i, line_idx1_j, line_idx2_i, line_idx2_j, line_idx3_i, line_idx3_j |
+                                            | line_idx0_i, line_idx0_j, line_idx1_i, line_idx1_j, line_idx2_i, line_idx2_j, line_idx3_i, line_idx3_j |
+                                            ...
+                                        '''
+                                        square_list.append(corner0[:2] + corner1[:2] + corner2[:2] + corner3[:2])
+                                        connect_list.append([corner0_line, corner1_line, corner2_line, corner3_line])
+                                        segments_list.append(corner0[2:] + corner1[2:] + corner2[2:] + corner3[2:])
+
+    def check_outside_inside(segments_info, connect_idx):
+        # return 'outside or inside', min distance, cover_param, peri_param
+        if connect_idx == segments_info[0]:
+            check_dist_mat = dist_inter_to_segment1
+        else:
+            check_dist_mat = dist_inter_to_segment2
+
+        i, j = segments_info
+        min_dist, max_dist = check_dist_mat[i, j, :]
+        connect_dist = dist_segments[connect_idx]
+        if max_dist > connect_dist:
+            return 'outside', min_dist, 0, 1
+        else:
+            return 'inside', min_dist, -1, -1
+
+    top_square = None
+
+    try:
+        map_size = input_shape[0] / 2
+        squares = np.array(square_list).reshape([-1, 4, 2])
+        score_array = []
+        connect_array = np.array(connect_list)
+        segments_array = np.array(segments_list).reshape([-1, 4, 2])
+
+        # get degree of corners:
+        squares_rollup = np.roll(squares, 1, axis=1)
+        squares_rolldown = np.roll(squares, -1, axis=1)
+        vec1 = squares_rollup - squares
+        normalized_vec1 = vec1 / (np.linalg.norm(vec1, axis=-1, keepdims=True) + 1e-10)
+        vec2 = squares_rolldown - squares
+        normalized_vec2 = vec2 / (np.linalg.norm(vec2, axis=-1, keepdims=True) + 1e-10)
+        inner_products = np.sum(normalized_vec1 * normalized_vec2, axis=-1)  # [n_squares, 4]
+        squares_degree = np.arccos(inner_products) * 180 / np.pi  # [n_squares, 4]
+
+        # get square score
+        overlap_scores = []
+        degree_scores = []
+        length_scores = []
+
+        for connects, segments, square, degree in zip(connect_array, segments_array, squares, squares_degree):
+            '''
+            0 -- 1
+            |    |
+            3 -- 2
+
+            # segments: [4, 2]
+            # connects: [4]
+            '''
+
+            ###################################### OVERLAP SCORES
+            cover = 0
+            perimeter = 0
+            # check 0 > 1 > 2 > 3
+            square_length = []
+
+            for start_idx in range(4):
+                end_idx = (start_idx + 1) % 4
+
+                connect_idx = connects[start_idx]  # segment idx of segment01
+                start_segments = segments[start_idx]
+                end_segments = segments[end_idx]
+
+                start_point = square[start_idx]
+                end_point = square[end_idx]
+
+                # check whether outside or inside
+                start_position, start_min, start_cover_param, start_peri_param = check_outside_inside(start_segments,
+                                                                                                      connect_idx)
+                end_position, end_min, end_cover_param, end_peri_param = check_outside_inside(end_segments, connect_idx)
+
+                cover += dist_segments[connect_idx] + start_cover_param * start_min + end_cover_param * end_min
+                perimeter += dist_segments[connect_idx] + start_peri_param * start_min + end_peri_param * end_min
+
+                square_length.append(
+                    dist_segments[connect_idx] + start_peri_param * start_min + end_peri_param * end_min)
+
+            overlap_scores.append(cover / perimeter)
+            ######################################
+            ###################################### DEGREE SCORES
+            '''
+            deg0 vs deg2
+            deg1 vs deg3
+            '''
+            deg0, deg1, deg2, deg3 = degree
+            deg_ratio1 = deg0 / deg2
+            if deg_ratio1 > 1.0:
+                deg_ratio1 = 1 / deg_ratio1
+            deg_ratio2 = deg1 / deg3
+            if deg_ratio2 > 1.0:
+                deg_ratio2 = 1 / deg_ratio2
+            degree_scores.append((deg_ratio1 + deg_ratio2) / 2)
+            ######################################
+            ###################################### LENGTH SCORES
+            '''
+            len0 vs len2
+            len1 vs len3
+            '''
+            len0, len1, len2, len3 = square_length
+            len_ratio1 = len0 / len2 if len2 > len0 else len2 / len0
+            len_ratio2 = len1 / len3 if len3 > len1 else len3 / len1
+            length_scores.append((len_ratio1 + len_ratio2) / 2)
+
+            ######################################
+
+        overlap_scores = np.array(overlap_scores)
+        overlap_scores /= np.max(overlap_scores)
+
+        degree_scores = np.array(degree_scores)
+        # degree_scores /= np.max(degree_scores)
+
+        length_scores = np.array(length_scores)
+
+        ###################################### AREA SCORES
+        area_scores = np.reshape(squares, [-1, 4, 2])
+        area_x = area_scores[:, :, 0]
+        area_y = area_scores[:, :, 1]
+        correction = area_x[:, -1] * area_y[:, 0] - area_y[:, -1] * area_x[:, 0]
+        area_scores = np.sum(area_x[:, :-1] * area_y[:, 1:], axis=-1) - np.sum(area_y[:, :-1] * area_x[:, 1:], axis=-1)
+        area_scores = 0.5 * np.abs(area_scores + correction)
+        area_scores /= (map_size * map_size)  # np.max(area_scores)
+        ######################################
+
+        ###################################### CENTER SCORES
+        centers = np.array([[256 // 2, 256 // 2]], dtype='float32')  # [1, 2]
+        # squares: [n, 4, 2]
+        square_centers = np.mean(squares, axis=1)  # [n, 2]
+        center2center = np.sqrt(np.sum((centers - square_centers) ** 2))
+        center_scores = center2center / (map_size / np.sqrt(2.0))
+
+        '''
+        score_w = [overlap, degree, area, center, length]
+        '''
+        score_w = [0.0, 1.0, 10.0, 0.5, 1.0]
+        score_array = params['w_overlap'] * overlap_scores \
+                      + params['w_degree'] * degree_scores \
+                      + params['w_area'] * area_scores \
+                      - params['w_center'] * center_scores \
+                      + params['w_length'] * length_scores
+
+        best_square = []
+
+        sorted_idx = np.argsort(score_array)[::-1]
+        score_array = score_array[sorted_idx]
+        squares = squares[sorted_idx]
+
+    except Exception as e:
+        pass
+
+    '''return list
+    merged_lines, squares, scores
+    '''
+
+    try:
+        new_segments[:, 0] = new_segments[:, 0] * 2 / input_shape[1] * original_shape[1]
+        new_segments[:, 1] = new_segments[:, 1] * 2 / input_shape[0] * original_shape[0]
+        new_segments[:, 2] = new_segments[:, 2] * 2 / input_shape[1] * original_shape[1]
+        new_segments[:, 3] = new_segments[:, 3] * 2 / input_shape[0] * original_shape[0]
+    except:
+        new_segments = []
+
+    try:
+        squares[:, :, 0] = squares[:, :, 0] * 2 / input_shape[1] * original_shape[1]
+        squares[:, :, 1] = squares[:, :, 1] * 2 / input_shape[0] * original_shape[0]
+    except:
+        squares = []
+        score_array = []
+
+    try:
+        inter_points = np.array(inter_points)
+        inter_points[:, 0] = inter_points[:, 0] * 2 / input_shape[1] * original_shape[1]
+        inter_points[:, 1] = inter_points[:, 1] * 2 / input_shape[0] * original_shape[0]
+    except:
+        inter_points = []
+
+    return new_segments, squares, score_array, inter_points
diff --git a/annotator/openpose/__init__.py b/annotator/openpose/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..8c26f1b37dae854f51da938da2fa67a8ef48ce5a
--- /dev/null
+++ b/annotator/openpose/__init__.py
@@ -0,0 +1,44 @@
+import os
+os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
+
+import torch
+import numpy as np
+from . import util
+from .body import Body
+from .hand import Hand
+from annotator.util import annotator_ckpts_path
+
+
+body_model_path = "https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/body_pose_model.pth"
+hand_model_path = "https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/hand_pose_model.pth"
+
+
+class OpenposeDetector:
+    def __init__(self):
+        body_modelpath = os.path.join(annotator_ckpts_path, "body_pose_model.pth")
+        hand_modelpath = os.path.join(annotator_ckpts_path, "hand_pose_model.pth")
+
+        if not os.path.exists(hand_modelpath):
+            from basicsr.utils.download_util import load_file_from_url
+            load_file_from_url(body_model_path, model_dir=annotator_ckpts_path)
+            load_file_from_url(hand_model_path, model_dir=annotator_ckpts_path)
+
+        self.body_estimation = Body(body_modelpath)
+        self.hand_estimation = Hand(hand_modelpath)
+
+    def __call__(self, oriImg, hand=False):
+        oriImg = oriImg[:, :, ::-1].copy()
+        with torch.no_grad():
+            candidate, subset = self.body_estimation(oriImg)
+            canvas = np.zeros_like(oriImg)
+            canvas = util.draw_bodypose(canvas, candidate, subset)
+            if hand:
+                hands_list = util.handDetect(candidate, subset, oriImg)
+                all_hand_peaks = []
+                for x, y, w, is_left in hands_list:
+                    peaks = self.hand_estimation(oriImg[y:y+w, x:x+w, :])
+                    peaks[:, 0] = np.where(peaks[:, 0] == 0, peaks[:, 0], peaks[:, 0] + x)
+                    peaks[:, 1] = np.where(peaks[:, 1] == 0, peaks[:, 1], peaks[:, 1] + y)
+                    all_hand_peaks.append(peaks)
+                canvas = util.draw_handpose(canvas, all_hand_peaks)
+            return canvas, dict(candidate=candidate.tolist(), subset=subset.tolist())
diff --git a/annotator/openpose/body.py b/annotator/openpose/body.py
new file mode 100644
index 0000000000000000000000000000000000000000..7c3cf7a388b4ac81004524e64125e383bdd455bd
--- /dev/null
+++ b/annotator/openpose/body.py
@@ -0,0 +1,219 @@
+import cv2
+import numpy as np
+import math
+import time
+from scipy.ndimage.filters import gaussian_filter
+import matplotlib.pyplot as plt
+import matplotlib
+import torch
+from torchvision import transforms
+
+from . import util
+from .model import bodypose_model
+
+class Body(object):
+    def __init__(self, model_path):
+        self.model = bodypose_model()
+        if torch.cuda.is_available():
+            self.model = self.model.cuda()
+            print('cuda')
+        model_dict = util.transfer(self.model, torch.load(model_path))
+        self.model.load_state_dict(model_dict)
+        self.model.eval()
+
+    def __call__(self, oriImg):
+        # scale_search = [0.5, 1.0, 1.5, 2.0]
+        scale_search = [0.5]
+        boxsize = 368
+        stride = 8
+        padValue = 128
+        thre1 = 0.1
+        thre2 = 0.05
+        multiplier = [x * boxsize / oriImg.shape[0] for x in scale_search]
+        heatmap_avg = np.zeros((oriImg.shape[0], oriImg.shape[1], 19))
+        paf_avg = np.zeros((oriImg.shape[0], oriImg.shape[1], 38))
+
+        for m in range(len(multiplier)):
+            scale = multiplier[m]
+            imageToTest = cv2.resize(oriImg, (0, 0), fx=scale, fy=scale, interpolation=cv2.INTER_CUBIC)
+            imageToTest_padded, pad = util.padRightDownCorner(imageToTest, stride, padValue)
+            im = np.transpose(np.float32(imageToTest_padded[:, :, :, np.newaxis]), (3, 2, 0, 1)) / 256 - 0.5
+            im = np.ascontiguousarray(im)
+
+            data = torch.from_numpy(im).float()
+            if torch.cuda.is_available():
+                data = data.cuda()
+            # data = data.permute([2, 0, 1]).unsqueeze(0).float()
+            with torch.no_grad():
+                Mconv7_stage6_L1, Mconv7_stage6_L2 = self.model(data)
+            Mconv7_stage6_L1 = Mconv7_stage6_L1.cpu().numpy()
+            Mconv7_stage6_L2 = Mconv7_stage6_L2.cpu().numpy()
+
+            # extract outputs, resize, and remove padding
+            # heatmap = np.transpose(np.squeeze(net.blobs[output_blobs.keys()[1]].data), (1, 2, 0))  # output 1 is heatmaps
+            heatmap = np.transpose(np.squeeze(Mconv7_stage6_L2), (1, 2, 0))  # output 1 is heatmaps
+            heatmap = cv2.resize(heatmap, (0, 0), fx=stride, fy=stride, interpolation=cv2.INTER_CUBIC)
+            heatmap = heatmap[:imageToTest_padded.shape[0] - pad[2], :imageToTest_padded.shape[1] - pad[3], :]
+            heatmap = cv2.resize(heatmap, (oriImg.shape[1], oriImg.shape[0]), interpolation=cv2.INTER_CUBIC)
+
+            # paf = np.transpose(np.squeeze(net.blobs[output_blobs.keys()[0]].data), (1, 2, 0))  # output 0 is PAFs
+            paf = np.transpose(np.squeeze(Mconv7_stage6_L1), (1, 2, 0))  # output 0 is PAFs
+            paf = cv2.resize(paf, (0, 0), fx=stride, fy=stride, interpolation=cv2.INTER_CUBIC)
+            paf = paf[:imageToTest_padded.shape[0] - pad[2], :imageToTest_padded.shape[1] - pad[3], :]
+            paf = cv2.resize(paf, (oriImg.shape[1], oriImg.shape[0]), interpolation=cv2.INTER_CUBIC)
+
+            heatmap_avg += heatmap_avg + heatmap / len(multiplier)
+            paf_avg += + paf / len(multiplier)
+
+        all_peaks = []
+        peak_counter = 0
+
+        for part in range(18):
+            map_ori = heatmap_avg[:, :, part]
+            one_heatmap = gaussian_filter(map_ori, sigma=3)
+
+            map_left = np.zeros(one_heatmap.shape)
+            map_left[1:, :] = one_heatmap[:-1, :]
+            map_right = np.zeros(one_heatmap.shape)
+            map_right[:-1, :] = one_heatmap[1:, :]
+            map_up = np.zeros(one_heatmap.shape)
+            map_up[:, 1:] = one_heatmap[:, :-1]
+            map_down = np.zeros(one_heatmap.shape)
+            map_down[:, :-1] = one_heatmap[:, 1:]
+
+            peaks_binary = np.logical_and.reduce(
+                (one_heatmap >= map_left, one_heatmap >= map_right, one_heatmap >= map_up, one_heatmap >= map_down, one_heatmap > thre1))
+            peaks = list(zip(np.nonzero(peaks_binary)[1], np.nonzero(peaks_binary)[0]))  # note reverse
+            peaks_with_score = [x + (map_ori[x[1], x[0]],) for x in peaks]
+            peak_id = range(peak_counter, peak_counter + len(peaks))
+            peaks_with_score_and_id = [peaks_with_score[i] + (peak_id[i],) for i in range(len(peak_id))]
+
+            all_peaks.append(peaks_with_score_and_id)
+            peak_counter += len(peaks)
+
+        # find connection in the specified sequence, center 29 is in the position 15
+        limbSeq = [[2, 3], [2, 6], [3, 4], [4, 5], [6, 7], [7, 8], [2, 9], [9, 10], \
+                   [10, 11], [2, 12], [12, 13], [13, 14], [2, 1], [1, 15], [15, 17], \
+                   [1, 16], [16, 18], [3, 17], [6, 18]]
+        # the middle joints heatmap correpondence
+        mapIdx = [[31, 32], [39, 40], [33, 34], [35, 36], [41, 42], [43, 44], [19, 20], [21, 22], \
+                  [23, 24], [25, 26], [27, 28], [29, 30], [47, 48], [49, 50], [53, 54], [51, 52], \
+                  [55, 56], [37, 38], [45, 46]]
+
+        connection_all = []
+        special_k = []
+        mid_num = 10
+
+        for k in range(len(mapIdx)):
+            score_mid = paf_avg[:, :, [x - 19 for x in mapIdx[k]]]
+            candA = all_peaks[limbSeq[k][0] - 1]
+            candB = all_peaks[limbSeq[k][1] - 1]
+            nA = len(candA)
+            nB = len(candB)
+            indexA, indexB = limbSeq[k]
+            if (nA != 0 and nB != 0):
+                connection_candidate = []
+                for i in range(nA):
+                    for j in range(nB):
+                        vec = np.subtract(candB[j][:2], candA[i][:2])
+                        norm = math.sqrt(vec[0] * vec[0] + vec[1] * vec[1])
+                        norm = max(0.001, norm)
+                        vec = np.divide(vec, norm)
+
+                        startend = list(zip(np.linspace(candA[i][0], candB[j][0], num=mid_num), \
+                                            np.linspace(candA[i][1], candB[j][1], num=mid_num)))
+
+                        vec_x = np.array([score_mid[int(round(startend[I][1])), int(round(startend[I][0])), 0] \
+                                          for I in range(len(startend))])
+                        vec_y = np.array([score_mid[int(round(startend[I][1])), int(round(startend[I][0])), 1] \
+                                          for I in range(len(startend))])
+
+                        score_midpts = np.multiply(vec_x, vec[0]) + np.multiply(vec_y, vec[1])
+                        score_with_dist_prior = sum(score_midpts) / len(score_midpts) + min(
+                            0.5 * oriImg.shape[0] / norm - 1, 0)
+                        criterion1 = len(np.nonzero(score_midpts > thre2)[0]) > 0.8 * len(score_midpts)
+                        criterion2 = score_with_dist_prior > 0
+                        if criterion1 and criterion2:
+                            connection_candidate.append(
+                                [i, j, score_with_dist_prior, score_with_dist_prior + candA[i][2] + candB[j][2]])
+
+                connection_candidate = sorted(connection_candidate, key=lambda x: x[2], reverse=True)
+                connection = np.zeros((0, 5))
+                for c in range(len(connection_candidate)):
+                    i, j, s = connection_candidate[c][0:3]
+                    if (i not in connection[:, 3] and j not in connection[:, 4]):
+                        connection = np.vstack([connection, [candA[i][3], candB[j][3], s, i, j]])
+                        if (len(connection) >= min(nA, nB)):
+                            break
+
+                connection_all.append(connection)
+            else:
+                special_k.append(k)
+                connection_all.append([])
+
+        # last number in each row is the total parts number of that person
+        # the second last number in each row is the score of the overall configuration
+        subset = -1 * np.ones((0, 20))
+        candidate = np.array([item for sublist in all_peaks for item in sublist])
+
+        for k in range(len(mapIdx)):
+            if k not in special_k:
+                partAs = connection_all[k][:, 0]
+                partBs = connection_all[k][:, 1]
+                indexA, indexB = np.array(limbSeq[k]) - 1
+
+                for i in range(len(connection_all[k])):  # = 1:size(temp,1)
+                    found = 0
+                    subset_idx = [-1, -1]
+                    for j in range(len(subset)):  # 1:size(subset,1):
+                        if subset[j][indexA] == partAs[i] or subset[j][indexB] == partBs[i]:
+                            subset_idx[found] = j
+                            found += 1
+
+                    if found == 1:
+                        j = subset_idx[0]
+                        if subset[j][indexB] != partBs[i]:
+                            subset[j][indexB] = partBs[i]
+                            subset[j][-1] += 1
+                            subset[j][-2] += candidate[partBs[i].astype(int), 2] + connection_all[k][i][2]
+                    elif found == 2:  # if found 2 and disjoint, merge them
+                        j1, j2 = subset_idx
+                        membership = ((subset[j1] >= 0).astype(int) + (subset[j2] >= 0).astype(int))[:-2]
+                        if len(np.nonzero(membership == 2)[0]) == 0:  # merge
+                            subset[j1][:-2] += (subset[j2][:-2] + 1)
+                            subset[j1][-2:] += subset[j2][-2:]
+                            subset[j1][-2] += connection_all[k][i][2]
+                            subset = np.delete(subset, j2, 0)
+                        else:  # as like found == 1
+                            subset[j1][indexB] = partBs[i]
+                            subset[j1][-1] += 1
+                            subset[j1][-2] += candidate[partBs[i].astype(int), 2] + connection_all[k][i][2]
+
+                    # if find no partA in the subset, create a new subset
+                    elif not found and k < 17:
+                        row = -1 * np.ones(20)
+                        row[indexA] = partAs[i]
+                        row[indexB] = partBs[i]
+                        row[-1] = 2
+                        row[-2] = sum(candidate[connection_all[k][i, :2].astype(int), 2]) + connection_all[k][i][2]
+                        subset = np.vstack([subset, row])
+        # delete some rows of subset which has few parts occur
+        deleteIdx = []
+        for i in range(len(subset)):
+            if subset[i][-1] < 4 or subset[i][-2] / subset[i][-1] < 0.4:
+                deleteIdx.append(i)
+        subset = np.delete(subset, deleteIdx, axis=0)
+
+        # subset: n*20 array, 0-17 is the index in candidate, 18 is the total score, 19 is the total parts
+        # candidate: x, y, score, id
+        return candidate, subset
+
+if __name__ == "__main__":
+    body_estimation = Body('../model/body_pose_model.pth')
+
+    test_image = '../images/ski.jpg'
+    oriImg = cv2.imread(test_image)  # B,G,R order
+    candidate, subset = body_estimation(oriImg)
+    canvas = util.draw_bodypose(oriImg, candidate, subset)
+    plt.imshow(canvas[:, :, [2, 1, 0]])
+    plt.show()
diff --git a/annotator/openpose/hand.py b/annotator/openpose/hand.py
new file mode 100644
index 0000000000000000000000000000000000000000..3d0bf17165ad7eb225332b51f4a2aa16718664b2
--- /dev/null
+++ b/annotator/openpose/hand.py
@@ -0,0 +1,86 @@
+import cv2
+import json
+import numpy as np
+import math
+import time
+from scipy.ndimage.filters import gaussian_filter
+import matplotlib.pyplot as plt
+import matplotlib
+import torch
+from skimage.measure import label
+
+from .model import handpose_model
+from . import util
+
+class Hand(object):
+    def __init__(self, model_path):
+        self.model = handpose_model()
+        if torch.cuda.is_available():
+            self.model = self.model.cuda()
+            print('cuda')
+        model_dict = util.transfer(self.model, torch.load(model_path))
+        self.model.load_state_dict(model_dict)
+        self.model.eval()
+
+    def __call__(self, oriImg):
+        scale_search = [0.5, 1.0, 1.5, 2.0]
+        # scale_search = [0.5]
+        boxsize = 368
+        stride = 8
+        padValue = 128
+        thre = 0.05
+        multiplier = [x * boxsize / oriImg.shape[0] for x in scale_search]
+        heatmap_avg = np.zeros((oriImg.shape[0], oriImg.shape[1], 22))
+        # paf_avg = np.zeros((oriImg.shape[0], oriImg.shape[1], 38))
+
+        for m in range(len(multiplier)):
+            scale = multiplier[m]
+            imageToTest = cv2.resize(oriImg, (0, 0), fx=scale, fy=scale, interpolation=cv2.INTER_CUBIC)
+            imageToTest_padded, pad = util.padRightDownCorner(imageToTest, stride, padValue)
+            im = np.transpose(np.float32(imageToTest_padded[:, :, :, np.newaxis]), (3, 2, 0, 1)) / 256 - 0.5
+            im = np.ascontiguousarray(im)
+
+            data = torch.from_numpy(im).float()
+            if torch.cuda.is_available():
+                data = data.cuda()
+            # data = data.permute([2, 0, 1]).unsqueeze(0).float()
+            with torch.no_grad():
+                output = self.model(data).cpu().numpy()
+                # output = self.model(data).numpy()q
+
+            # extract outputs, resize, and remove padding
+            heatmap = np.transpose(np.squeeze(output), (1, 2, 0))  # output 1 is heatmaps
+            heatmap = cv2.resize(heatmap, (0, 0), fx=stride, fy=stride, interpolation=cv2.INTER_CUBIC)
+            heatmap = heatmap[:imageToTest_padded.shape[0] - pad[2], :imageToTest_padded.shape[1] - pad[3], :]
+            heatmap = cv2.resize(heatmap, (oriImg.shape[1], oriImg.shape[0]), interpolation=cv2.INTER_CUBIC)
+
+            heatmap_avg += heatmap / len(multiplier)
+
+        all_peaks = []
+        for part in range(21):
+            map_ori = heatmap_avg[:, :, part]
+            one_heatmap = gaussian_filter(map_ori, sigma=3)
+            binary = np.ascontiguousarray(one_heatmap > thre, dtype=np.uint8)
+            # 全部小于阈值
+            if np.sum(binary) == 0:
+                all_peaks.append([0, 0])
+                continue
+            label_img, label_numbers = label(binary, return_num=True, connectivity=binary.ndim)
+            max_index = np.argmax([np.sum(map_ori[label_img == i]) for i in range(1, label_numbers + 1)]) + 1
+            label_img[label_img != max_index] = 0
+            map_ori[label_img == 0] = 0
+
+            y, x = util.npmax(map_ori)
+            all_peaks.append([x, y])
+        return np.array(all_peaks)
+
+if __name__ == "__main__":
+    hand_estimation = Hand('../model/hand_pose_model.pth')
+
+    # test_image = '../images/hand.jpg'
+    test_image = '../images/hand.jpg'
+    oriImg = cv2.imread(test_image)  # B,G,R order
+    peaks = hand_estimation(oriImg)
+    canvas = util.draw_handpose(oriImg, peaks, True)
+    cv2.imshow('', canvas)
+    cv2.waitKey(0)
\ No newline at end of file
diff --git a/annotator/openpose/model.py b/annotator/openpose/model.py
new file mode 100644
index 0000000000000000000000000000000000000000..5dfc80de827a17beccb9b0f3f7588545be78c9de
--- /dev/null
+++ b/annotator/openpose/model.py
@@ -0,0 +1,219 @@
+import torch
+from collections import OrderedDict
+
+import torch
+import torch.nn as nn
+
+def make_layers(block, no_relu_layers):
+    layers = []
+    for layer_name, v in block.items():
+        if 'pool' in layer_name:
+            layer = nn.MaxPool2d(kernel_size=v[0], stride=v[1],
+                                    padding=v[2])
+            layers.append((layer_name, layer))
+        else:
+            conv2d = nn.Conv2d(in_channels=v[0], out_channels=v[1],
+                               kernel_size=v[2], stride=v[3],
+                               padding=v[4])
+            layers.append((layer_name, conv2d))
+            if layer_name not in no_relu_layers:
+                layers.append(('relu_'+layer_name, nn.ReLU(inplace=True)))
+
+    return nn.Sequential(OrderedDict(layers))
+
+class bodypose_model(nn.Module):
+    def __init__(self):
+        super(bodypose_model, self).__init__()
+
+        # these layers have no relu layer
+        no_relu_layers = ['conv5_5_CPM_L1', 'conv5_5_CPM_L2', 'Mconv7_stage2_L1',\
+                          'Mconv7_stage2_L2', 'Mconv7_stage3_L1', 'Mconv7_stage3_L2',\
+                          'Mconv7_stage4_L1', 'Mconv7_stage4_L2', 'Mconv7_stage5_L1',\
+                          'Mconv7_stage5_L2', 'Mconv7_stage6_L1', 'Mconv7_stage6_L1']
+        blocks = {}
+        block0 = OrderedDict([
+                      ('conv1_1', [3, 64, 3, 1, 1]),
+                      ('conv1_2', [64, 64, 3, 1, 1]),
+                      ('pool1_stage1', [2, 2, 0]),
+                      ('conv2_1', [64, 128, 3, 1, 1]),
+                      ('conv2_2', [128, 128, 3, 1, 1]),
+                      ('pool2_stage1', [2, 2, 0]),
+                      ('conv3_1', [128, 256, 3, 1, 1]),
+                      ('conv3_2', [256, 256, 3, 1, 1]),
+                      ('conv3_3', [256, 256, 3, 1, 1]),
+                      ('conv3_4', [256, 256, 3, 1, 1]),
+                      ('pool3_stage1', [2, 2, 0]),
+                      ('conv4_1', [256, 512, 3, 1, 1]),
+                      ('conv4_2', [512, 512, 3, 1, 1]),
+                      ('conv4_3_CPM', [512, 256, 3, 1, 1]),
+                      ('conv4_4_CPM', [256, 128, 3, 1, 1])
+                  ])
+
+
+        # Stage 1
+        block1_1 = OrderedDict([
+                        ('conv5_1_CPM_L1', [128, 128, 3, 1, 1]),
+                        ('conv5_2_CPM_L1', [128, 128, 3, 1, 1]),
+                        ('conv5_3_CPM_L1', [128, 128, 3, 1, 1]),
+                        ('conv5_4_CPM_L1', [128, 512, 1, 1, 0]),
+                        ('conv5_5_CPM_L1', [512, 38, 1, 1, 0])
+                    ])
+
+        block1_2 = OrderedDict([
+                        ('conv5_1_CPM_L2', [128, 128, 3, 1, 1]),
+                        ('conv5_2_CPM_L2', [128, 128, 3, 1, 1]),
+                        ('conv5_3_CPM_L2', [128, 128, 3, 1, 1]),
+                        ('conv5_4_CPM_L2', [128, 512, 1, 1, 0]),
+                        ('conv5_5_CPM_L2', [512, 19, 1, 1, 0])
+                    ])
+        blocks['block1_1'] = block1_1
+        blocks['block1_2'] = block1_2
+
+        self.model0 = make_layers(block0, no_relu_layers)
+
+        # Stages 2 - 6
+        for i in range(2, 7):
+            blocks['block%d_1' % i] = OrderedDict([
+                    ('Mconv1_stage%d_L1' % i, [185, 128, 7, 1, 3]),
+                    ('Mconv2_stage%d_L1' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv3_stage%d_L1' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv4_stage%d_L1' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv5_stage%d_L1' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv6_stage%d_L1' % i, [128, 128, 1, 1, 0]),
+                    ('Mconv7_stage%d_L1' % i, [128, 38, 1, 1, 0])
+                ])
+
+            blocks['block%d_2' % i] = OrderedDict([
+                    ('Mconv1_stage%d_L2' % i, [185, 128, 7, 1, 3]),
+                    ('Mconv2_stage%d_L2' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv3_stage%d_L2' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv4_stage%d_L2' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv5_stage%d_L2' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv6_stage%d_L2' % i, [128, 128, 1, 1, 0]),
+                    ('Mconv7_stage%d_L2' % i, [128, 19, 1, 1, 0])
+                ])
+
+        for k in blocks.keys():
+            blocks[k] = make_layers(blocks[k], no_relu_layers)
+
+        self.model1_1 = blocks['block1_1']
+        self.model2_1 = blocks['block2_1']
+        self.model3_1 = blocks['block3_1']
+        self.model4_1 = blocks['block4_1']
+        self.model5_1 = blocks['block5_1']
+        self.model6_1 = blocks['block6_1']
+
+        self.model1_2 = blocks['block1_2']
+        self.model2_2 = blocks['block2_2']
+        self.model3_2 = blocks['block3_2']
+        self.model4_2 = blocks['block4_2']
+        self.model5_2 = blocks['block5_2']
+        self.model6_2 = blocks['block6_2']
+
+
+    def forward(self, x):
+
+        out1 = self.model0(x)
+
+        out1_1 = self.model1_1(out1)
+        out1_2 = self.model1_2(out1)
+        out2 = torch.cat([out1_1, out1_2, out1], 1)
+
+        out2_1 = self.model2_1(out2)
+        out2_2 = self.model2_2(out2)
+        out3 = torch.cat([out2_1, out2_2, out1], 1)
+
+        out3_1 = self.model3_1(out3)
+        out3_2 = self.model3_2(out3)
+        out4 = torch.cat([out3_1, out3_2, out1], 1)
+
+        out4_1 = self.model4_1(out4)
+        out4_2 = self.model4_2(out4)
+        out5 = torch.cat([out4_1, out4_2, out1], 1)
+
+        out5_1 = self.model5_1(out5)
+        out5_2 = self.model5_2(out5)
+        out6 = torch.cat([out5_1, out5_2, out1], 1)
+
+        out6_1 = self.model6_1(out6)
+        out6_2 = self.model6_2(out6)
+
+        return out6_1, out6_2
+
+class handpose_model(nn.Module):
+    def __init__(self):
+        super(handpose_model, self).__init__()
+
+        # these layers have no relu layer
+        no_relu_layers = ['conv6_2_CPM', 'Mconv7_stage2', 'Mconv7_stage3',\
+                          'Mconv7_stage4', 'Mconv7_stage5', 'Mconv7_stage6']
+        # stage 1
+        block1_0 = OrderedDict([
+                ('conv1_1', [3, 64, 3, 1, 1]),
+                ('conv1_2', [64, 64, 3, 1, 1]),
+                ('pool1_stage1', [2, 2, 0]),
+                ('conv2_1', [64, 128, 3, 1, 1]),
+                ('conv2_2', [128, 128, 3, 1, 1]),
+                ('pool2_stage1', [2, 2, 0]),
+                ('conv3_1', [128, 256, 3, 1, 1]),
+                ('conv3_2', [256, 256, 3, 1, 1]),
+                ('conv3_3', [256, 256, 3, 1, 1]),
+                ('conv3_4', [256, 256, 3, 1, 1]),
+                ('pool3_stage1', [2, 2, 0]),
+                ('conv4_1', [256, 512, 3, 1, 1]),
+                ('conv4_2', [512, 512, 3, 1, 1]),
+                ('conv4_3', [512, 512, 3, 1, 1]),
+                ('conv4_4', [512, 512, 3, 1, 1]),
+                ('conv5_1', [512, 512, 3, 1, 1]),
+                ('conv5_2', [512, 512, 3, 1, 1]),
+                ('conv5_3_CPM', [512, 128, 3, 1, 1])
+            ])
+
+        block1_1 = OrderedDict([
+            ('conv6_1_CPM', [128, 512, 1, 1, 0]),
+            ('conv6_2_CPM', [512, 22, 1, 1, 0])
+        ])
+
+        blocks = {}
+        blocks['block1_0'] = block1_0
+        blocks['block1_1'] = block1_1
+
+        # stage 2-6
+        for i in range(2, 7):
+            blocks['block%d' % i] = OrderedDict([
+                    ('Mconv1_stage%d' % i, [150, 128, 7, 1, 3]),
+                    ('Mconv2_stage%d' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv3_stage%d' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv4_stage%d' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv5_stage%d' % i, [128, 128, 7, 1, 3]),
+                    ('Mconv6_stage%d' % i, [128, 128, 1, 1, 0]),
+                    ('Mconv7_stage%d' % i, [128, 22, 1, 1, 0])
+                ])
+
+        for k in blocks.keys():
+            blocks[k] = make_layers(blocks[k], no_relu_layers)
+
+        self.model1_0 = blocks['block1_0']
+        self.model1_1 = blocks['block1_1']
+        self.model2 = blocks['block2']
+        self.model3 = blocks['block3']
+        self.model4 = blocks['block4']
+        self.model5 = blocks['block5']
+        self.model6 = blocks['block6']
+
+    def forward(self, x):
+        out1_0 = self.model1_0(x)
+        out1_1 = self.model1_1(out1_0)
+        concat_stage2 = torch.cat([out1_1, out1_0], 1)
+        out_stage2 = self.model2(concat_stage2)
+        concat_stage3 = torch.cat([out_stage2, out1_0], 1)
+        out_stage3 = self.model3(concat_stage3)
+        concat_stage4 = torch.cat([out_stage3, out1_0], 1)
+        out_stage4 = self.model4(concat_stage4)
+        concat_stage5 = torch.cat([out_stage4, out1_0], 1)
+        out_stage5 = self.model5(concat_stage5)
+        concat_stage6 = torch.cat([out_stage5, out1_0], 1)
+        out_stage6 = self.model6(concat_stage6)
+        return out_stage6
+
+
diff --git a/annotator/openpose/util.py b/annotator/openpose/util.py
new file mode 100644
index 0000000000000000000000000000000000000000..6f91ae0e65abaf0cbd62d803f56498991141e61b
--- /dev/null
+++ b/annotator/openpose/util.py
@@ -0,0 +1,164 @@
+import math
+import numpy as np
+import matplotlib
+import cv2
+
+
+def padRightDownCorner(img, stride, padValue):
+    h = img.shape[0]
+    w = img.shape[1]
+
+    pad = 4 * [None]
+    pad[0] = 0 # up
+    pad[1] = 0 # left
+    pad[2] = 0 if (h % stride == 0) else stride - (h % stride) # down
+    pad[3] = 0 if (w % stride == 0) else stride - (w % stride) # right
+
+    img_padded = img
+    pad_up = np.tile(img_padded[0:1, :, :]*0 + padValue, (pad[0], 1, 1))
+    img_padded = np.concatenate((pad_up, img_padded), axis=0)
+    pad_left = np.tile(img_padded[:, 0:1, :]*0 + padValue, (1, pad[1], 1))
+    img_padded = np.concatenate((pad_left, img_padded), axis=1)
+    pad_down = np.tile(img_padded[-2:-1, :, :]*0 + padValue, (pad[2], 1, 1))
+    img_padded = np.concatenate((img_padded, pad_down), axis=0)
+    pad_right = np.tile(img_padded[:, -2:-1, :]*0 + padValue, (1, pad[3], 1))
+    img_padded = np.concatenate((img_padded, pad_right), axis=1)
+
+    return img_padded, pad
+
+# transfer caffe model to pytorch which will match the layer name
+def transfer(model, model_weights):
+    transfered_model_weights = {}
+    for weights_name in model.state_dict().keys():
+        transfered_model_weights[weights_name] = model_weights['.'.join(weights_name.split('.')[1:])]
+    return transfered_model_weights
+
+# draw the body keypoint and lims
+def draw_bodypose(canvas, candidate, subset):
+    stickwidth = 4
+    limbSeq = [[2, 3], [2, 6], [3, 4], [4, 5], [6, 7], [7, 8], [2, 9], [9, 10], \
+               [10, 11], [2, 12], [12, 13], [13, 14], [2, 1], [1, 15], [15, 17], \
+               [1, 16], [16, 18], [3, 17], [6, 18]]
+
+    colors = [[255, 0, 0], [255, 85, 0], [255, 170, 0], [255, 255, 0], [170, 255, 0], [85, 255, 0], [0, 255, 0], \
+              [0, 255, 85], [0, 255, 170], [0, 255, 255], [0, 170, 255], [0, 85, 255], [0, 0, 255], [85, 0, 255], \
+              [170, 0, 255], [255, 0, 255], [255, 0, 170], [255, 0, 85]]
+    for i in range(18):
+        for n in range(len(subset)):
+            index = int(subset[n][i])
+            if index == -1:
+                continue
+            x, y = candidate[index][0:2]
+            cv2.circle(canvas, (int(x), int(y)), 4, colors[i], thickness=-1)
+    for i in range(17):
+        for n in range(len(subset)):
+            index = subset[n][np.array(limbSeq[i]) - 1]
+            if -1 in index:
+                continue
+            cur_canvas = canvas.copy()
+            Y = candidate[index.astype(int), 0]
+            X = candidate[index.astype(int), 1]
+            mX = np.mean(X)
+            mY = np.mean(Y)
+            length = ((X[0] - X[1]) ** 2 + (Y[0] - Y[1]) ** 2) ** 0.5
+            angle = math.degrees(math.atan2(X[0] - X[1], Y[0] - Y[1]))
+            polygon = cv2.ellipse2Poly((int(mY), int(mX)), (int(length / 2), stickwidth), int(angle), 0, 360, 1)
+            cv2.fillConvexPoly(cur_canvas, polygon, colors[i])
+            canvas = cv2.addWeighted(canvas, 0.4, cur_canvas, 0.6, 0)
+    # plt.imsave("preview.jpg", canvas[:, :, [2, 1, 0]])
+    # plt.imshow(canvas[:, :, [2, 1, 0]])
+    return canvas
+
+
+# image drawed by opencv is not good.
+def draw_handpose(canvas, all_hand_peaks, show_number=False):
+    edges = [[0, 1], [1, 2], [2, 3], [3, 4], [0, 5], [5, 6], [6, 7], [7, 8], [0, 9], [9, 10], \
+             [10, 11], [11, 12], [0, 13], [13, 14], [14, 15], [15, 16], [0, 17], [17, 18], [18, 19], [19, 20]]
+
+    for peaks in all_hand_peaks:
+        for ie, e in enumerate(edges):
+            if np.sum(np.all(peaks[e], axis=1)==0)==0:
+                x1, y1 = peaks[e[0]]
+                x2, y2 = peaks[e[1]]
+                cv2.line(canvas, (x1, y1), (x2, y2), matplotlib.colors.hsv_to_rgb([ie/float(len(edges)), 1.0, 1.0])*255, thickness=2)
+
+        for i, keyponit in enumerate(peaks):
+            x, y = keyponit
+            cv2.circle(canvas, (x, y), 4, (0, 0, 255), thickness=-1)
+            if show_number:
+                cv2.putText(canvas, str(i), (x, y), cv2.FONT_HERSHEY_SIMPLEX, 0.3, (0, 0, 0), lineType=cv2.LINE_AA)
+    return canvas
+
+# detect hand according to body pose keypoints
+# please refer to https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/src/openpose/hand/handDetector.cpp
+def handDetect(candidate, subset, oriImg):
+    # right hand: wrist 4, elbow 3, shoulder 2
+    # left hand: wrist 7, elbow 6, shoulder 5
+    ratioWristElbow = 0.33
+    detect_result = []
+    image_height, image_width = oriImg.shape[0:2]
+    for person in subset.astype(int):
+        # if any of three not detected
+        has_left = np.sum(person[[5, 6, 7]] == -1) == 0
+        has_right = np.sum(person[[2, 3, 4]] == -1) == 0
+        if not (has_left or has_right):
+            continue
+        hands = []
+        #left hand
+        if has_left:
+            left_shoulder_index, left_elbow_index, left_wrist_index = person[[5, 6, 7]]
+            x1, y1 = candidate[left_shoulder_index][:2]
+            x2, y2 = candidate[left_elbow_index][:2]
+            x3, y3 = candidate[left_wrist_index][:2]
+            hands.append([x1, y1, x2, y2, x3, y3, True])
+        # right hand
+        if has_right:
+            right_shoulder_index, right_elbow_index, right_wrist_index = person[[2, 3, 4]]
+            x1, y1 = candidate[right_shoulder_index][:2]
+            x2, y2 = candidate[right_elbow_index][:2]
+            x3, y3 = candidate[right_wrist_index][:2]
+            hands.append([x1, y1, x2, y2, x3, y3, False])
+
+        for x1, y1, x2, y2, x3, y3, is_left in hands:
+            # pos_hand = pos_wrist + ratio * (pos_wrist - pos_elbox) = (1 + ratio) * pos_wrist - ratio * pos_elbox
+            # handRectangle.x = posePtr[wrist*3] + ratioWristElbow * (posePtr[wrist*3] - posePtr[elbow*3]);
+            # handRectangle.y = posePtr[wrist*3+1] + ratioWristElbow * (posePtr[wrist*3+1] - posePtr[elbow*3+1]);
+            # const auto distanceWristElbow = getDistance(poseKeypoints, person, wrist, elbow);
+            # const auto distanceElbowShoulder = getDistance(poseKeypoints, person, elbow, shoulder);
+            # handRectangle.width = 1.5f * fastMax(distanceWristElbow, 0.9f * distanceElbowShoulder);
+            x = x3 + ratioWristElbow * (x3 - x2)
+            y = y3 + ratioWristElbow * (y3 - y2)
+            distanceWristElbow = math.sqrt((x3 - x2) ** 2 + (y3 - y2) ** 2)
+            distanceElbowShoulder = math.sqrt((x2 - x1) ** 2 + (y2 - y1) ** 2)
+            width = 1.5 * max(distanceWristElbow, 0.9 * distanceElbowShoulder)
+            # x-y refers to the center --> offset to topLeft point
+            # handRectangle.x -= handRectangle.width / 2.f;
+            # handRectangle.y -= handRectangle.height / 2.f;
+            x -= width / 2
+            y -= width / 2  # width = height
+            # overflow the image
+            if x < 0: x = 0
+            if y < 0: y = 0
+            width1 = width
+            width2 = width
+            if x + width > image_width: width1 = image_width - x
+            if y + width > image_height: width2 = image_height - y
+            width = min(width1, width2)
+            # the max hand box value is 20 pixels
+            if width >= 20:
+                detect_result.append([int(x), int(y), int(width), is_left])
+
+    '''
+    return value: [[x, y, w, True if left hand else False]].
+    width=height since the network require squared input.
+    x, y is the coordinate of top left 
+    '''
+    return detect_result
+
+# get max index of 2d array
+def npmax(array):
+    arrayindex = array.argmax(1)
+    arrayvalue = array.max(1)
+    i = arrayvalue.argmax()
+    j = arrayindex[i]
+    return i, j
diff --git a/annotator/uniformer/__init__.py b/annotator/uniformer/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..6be429542e4908c2b7648e7ee7c9c5f8253e7c94
--- /dev/null
+++ b/annotator/uniformer/__init__.py
@@ -0,0 +1,23 @@
+import os
+
+from annotator.uniformer.mmseg.apis import init_segmentor, inference_segmentor, show_result_pyplot
+from annotator.uniformer.mmseg.core.evaluation import get_palette
+from annotator.util import annotator_ckpts_path
+
+
+checkpoint_file = "https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/upernet_global_small.pth"
+
+
+class UniformerDetector:
+    def __init__(self):
+        modelpath = os.path.join(annotator_ckpts_path, "upernet_global_small.pth")
+        if not os.path.exists(modelpath):
+            from basicsr.utils.download_util import load_file_from_url
+            load_file_from_url(checkpoint_file, model_dir=annotator_ckpts_path)
+        config_file = os.path.join(os.path.dirname(annotator_ckpts_path), "uniformer", "exp", "upernet_global_small", "config.py")
+        self.model = init_segmentor(config_file, modelpath).cuda()
+
+    def __call__(self, img):
+        result = inference_segmentor(self.model, img)
+        res_img = show_result_pyplot(self.model, img, result, get_palette('ade'), opacity=1)
+        return res_img
diff --git a/annotator/uniformer/configs/_base_/datasets/ade20k.py b/annotator/uniformer/configs/_base_/datasets/ade20k.py
new file mode 100644
index 0000000000000000000000000000000000000000..efc8b4bb20c981f3db6df7eb52b3dc0744c94cc0
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/ade20k.py
@@ -0,0 +1,54 @@
+# dataset settings
+dataset_type = 'ADE20KDataset'
+data_root = 'data/ade/ADEChallengeData2016'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+crop_size = (512, 512)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', reduce_zero_label=True),
+    dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg']),
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(2048, 512),
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img']),
+        ])
+]
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/training',
+        ann_dir='annotations/training',
+        pipeline=train_pipeline),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/chase_db1.py b/annotator/uniformer/configs/_base_/datasets/chase_db1.py
new file mode 100644
index 0000000000000000000000000000000000000000..298594ea925f87f22b37094a2ec50e370aec96a0
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/chase_db1.py
@@ -0,0 +1,59 @@
+# dataset settings
+dataset_type = 'ChaseDB1Dataset'
+data_root = 'data/CHASE_DB1'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+img_scale = (960, 999)
+crop_size = (128, 128)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=img_scale, ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg'])
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=img_scale,
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75, 2.0],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type='RepeatDataset',
+        times=40000,
+        dataset=dict(
+            type=dataset_type,
+            data_root=data_root,
+            img_dir='images/training',
+            ann_dir='annotations/training',
+            pipeline=train_pipeline)),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/cityscapes.py b/annotator/uniformer/configs/_base_/datasets/cityscapes.py
new file mode 100644
index 0000000000000000000000000000000000000000..f21867c63e1835f6fceb61f066e802fd8fd2a735
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/cityscapes.py
@@ -0,0 +1,54 @@
+# dataset settings
+dataset_type = 'CityscapesDataset'
+data_root = 'data/cityscapes/'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+crop_size = (512, 1024)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=(2048, 1024), ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg']),
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(2048, 1024),
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img']),
+        ])
+]
+data = dict(
+    samples_per_gpu=2,
+    workers_per_gpu=2,
+    train=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='leftImg8bit/train',
+        ann_dir='gtFine/train',
+        pipeline=train_pipeline),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='leftImg8bit/val',
+        ann_dir='gtFine/val',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='leftImg8bit/val',
+        ann_dir='gtFine/val',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/cityscapes_769x769.py b/annotator/uniformer/configs/_base_/datasets/cityscapes_769x769.py
new file mode 100644
index 0000000000000000000000000000000000000000..336c7b254fe392b4703039fec86a83acdbd2e1a5
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/cityscapes_769x769.py
@@ -0,0 +1,35 @@
+_base_ = './cityscapes.py'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+crop_size = (769, 769)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=(2049, 1025), ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg']),
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(2049, 1025),
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img']),
+        ])
+]
+data = dict(
+    train=dict(pipeline=train_pipeline),
+    val=dict(pipeline=test_pipeline),
+    test=dict(pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/drive.py b/annotator/uniformer/configs/_base_/datasets/drive.py
new file mode 100644
index 0000000000000000000000000000000000000000..06e8ff606e0d2a4514ec8b7d2c6c436a32efcbf4
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/drive.py
@@ -0,0 +1,59 @@
+# dataset settings
+dataset_type = 'DRIVEDataset'
+data_root = 'data/DRIVE'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+img_scale = (584, 565)
+crop_size = (64, 64)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=img_scale, ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg'])
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=img_scale,
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75, 2.0],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type='RepeatDataset',
+        times=40000,
+        dataset=dict(
+            type=dataset_type,
+            data_root=data_root,
+            img_dir='images/training',
+            ann_dir='annotations/training',
+            pipeline=train_pipeline)),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/hrf.py b/annotator/uniformer/configs/_base_/datasets/hrf.py
new file mode 100644
index 0000000000000000000000000000000000000000..242d790eb1b83e75cf6b7eaa7a35c674099311ad
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/hrf.py
@@ -0,0 +1,59 @@
+# dataset settings
+dataset_type = 'HRFDataset'
+data_root = 'data/HRF'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+img_scale = (2336, 3504)
+crop_size = (256, 256)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=img_scale, ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg'])
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=img_scale,
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75, 2.0],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type='RepeatDataset',
+        times=40000,
+        dataset=dict(
+            type=dataset_type,
+            data_root=data_root,
+            img_dir='images/training',
+            ann_dir='annotations/training',
+            pipeline=train_pipeline)),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/pascal_context.py b/annotator/uniformer/configs/_base_/datasets/pascal_context.py
new file mode 100644
index 0000000000000000000000000000000000000000..ff65bad1b86d7e3a5980bb5b9fc55798dc8df5f4
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/pascal_context.py
@@ -0,0 +1,60 @@
+# dataset settings
+dataset_type = 'PascalContextDataset'
+data_root = 'data/VOCdevkit/VOC2010/'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+
+img_scale = (520, 520)
+crop_size = (480, 480)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=img_scale, ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg']),
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=img_scale,
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img']),
+        ])
+]
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClassContext',
+        split='ImageSets/SegmentationContext/train.txt',
+        pipeline=train_pipeline),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClassContext',
+        split='ImageSets/SegmentationContext/val.txt',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClassContext',
+        split='ImageSets/SegmentationContext/val.txt',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/pascal_context_59.py b/annotator/uniformer/configs/_base_/datasets/pascal_context_59.py
new file mode 100644
index 0000000000000000000000000000000000000000..37585abab89834b95cd5bdd993b994fca1db65f6
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/pascal_context_59.py
@@ -0,0 +1,60 @@
+# dataset settings
+dataset_type = 'PascalContextDataset59'
+data_root = 'data/VOCdevkit/VOC2010/'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+
+img_scale = (520, 520)
+crop_size = (480, 480)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations', reduce_zero_label=True),
+    dict(type='Resize', img_scale=img_scale, ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg']),
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=img_scale,
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img']),
+        ])
+]
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClassContext',
+        split='ImageSets/SegmentationContext/train.txt',
+        pipeline=train_pipeline),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClassContext',
+        split='ImageSets/SegmentationContext/val.txt',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClassContext',
+        split='ImageSets/SegmentationContext/val.txt',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/pascal_voc12.py b/annotator/uniformer/configs/_base_/datasets/pascal_voc12.py
new file mode 100644
index 0000000000000000000000000000000000000000..ba1d42d0c5781f56dc177d860d856bb34adce555
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/pascal_voc12.py
@@ -0,0 +1,57 @@
+# dataset settings
+dataset_type = 'PascalVOCDataset'
+data_root = 'data/VOCdevkit/VOC2012'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+crop_size = (512, 512)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg']),
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=(2048, 512),
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img']),
+        ])
+]
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClass',
+        split='ImageSets/Segmentation/train.txt',
+        pipeline=train_pipeline),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClass',
+        split='ImageSets/Segmentation/val.txt',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='JPEGImages',
+        ann_dir='SegmentationClass',
+        split='ImageSets/Segmentation/val.txt',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/datasets/pascal_voc12_aug.py b/annotator/uniformer/configs/_base_/datasets/pascal_voc12_aug.py
new file mode 100644
index 0000000000000000000000000000000000000000..3f23b6717d53ad29f02dd15046802a2631a5076b
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/pascal_voc12_aug.py
@@ -0,0 +1,9 @@
+_base_ = './pascal_voc12.py'
+# dataset settings
+data = dict(
+    train=dict(
+        ann_dir=['SegmentationClass', 'SegmentationClassAug'],
+        split=[
+            'ImageSets/Segmentation/train.txt',
+            'ImageSets/Segmentation/aug.txt'
+        ]))
diff --git a/annotator/uniformer/configs/_base_/datasets/stare.py b/annotator/uniformer/configs/_base_/datasets/stare.py
new file mode 100644
index 0000000000000000000000000000000000000000..3f71b25488cc11a6b4d582ac52b5a24e1ad1cf8e
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/datasets/stare.py
@@ -0,0 +1,59 @@
+# dataset settings
+dataset_type = 'STAREDataset'
+data_root = 'data/STARE'
+img_norm_cfg = dict(
+    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+img_scale = (605, 700)
+crop_size = (128, 128)
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='LoadAnnotations'),
+    dict(type='Resize', img_scale=img_scale, ratio_range=(0.5, 2.0)),
+    dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='PhotoMetricDistortion'),
+    dict(type='Normalize', **img_norm_cfg),
+    dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
+    dict(type='DefaultFormatBundle'),
+    dict(type='Collect', keys=['img', 'gt_semantic_seg'])
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='MultiScaleFlipAug',
+        img_scale=img_scale,
+        # img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75, 2.0],
+        flip=False,
+        transforms=[
+            dict(type='Resize', keep_ratio=True),
+            dict(type='RandomFlip'),
+            dict(type='Normalize', **img_norm_cfg),
+            dict(type='ImageToTensor', keys=['img']),
+            dict(type='Collect', keys=['img'])
+        ])
+]
+
+data = dict(
+    samples_per_gpu=4,
+    workers_per_gpu=4,
+    train=dict(
+        type='RepeatDataset',
+        times=40000,
+        dataset=dict(
+            type=dataset_type,
+            data_root=data_root,
+            img_dir='images/training',
+            ann_dir='annotations/training',
+            pipeline=train_pipeline)),
+    val=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline),
+    test=dict(
+        type=dataset_type,
+        data_root=data_root,
+        img_dir='images/validation',
+        ann_dir='annotations/validation',
+        pipeline=test_pipeline))
diff --git a/annotator/uniformer/configs/_base_/default_runtime.py b/annotator/uniformer/configs/_base_/default_runtime.py
new file mode 100644
index 0000000000000000000000000000000000000000..b564cc4e7e7d9a67dacaaddecb100e4d8f5c005b
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/default_runtime.py
@@ -0,0 +1,14 @@
+# yapf:disable
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook', by_epoch=False),
+        # dict(type='TensorboardLoggerHook')
+    ])
+# yapf:enable
+dist_params = dict(backend='nccl')
+log_level = 'INFO'
+load_from = None
+resume_from = None
+workflow = [('train', 1)]
+cudnn_benchmark = True
diff --git a/annotator/uniformer/configs/_base_/models/ann_r50-d8.py b/annotator/uniformer/configs/_base_/models/ann_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..a2cb653827e44e6015b3b83bc578003e614a6aa1
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/ann_r50-d8.py
@@ -0,0 +1,46 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='ANNHead',
+        in_channels=[1024, 2048],
+        in_index=[2, 3],
+        channels=512,
+        project_channels=256,
+        query_scales=(1, ),
+        key_pool_scales=(1, 3, 6, 8),
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/apcnet_r50-d8.py b/annotator/uniformer/configs/_base_/models/apcnet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..c8f5316cbcf3896ba9de7ca2c801eba512f01d5e
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/apcnet_r50-d8.py
@@ -0,0 +1,44 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='APCHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        pool_scales=(1, 2, 3, 6),
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=dict(type='SyncBN', requires_grad=True),
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/ccnet_r50-d8.py b/annotator/uniformer/configs/_base_/models/ccnet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..794148f576b9e215c3c6963e73dffe98204b7717
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/ccnet_r50-d8.py
@@ -0,0 +1,44 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='CCHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        recurrence=2,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/cgnet.py b/annotator/uniformer/configs/_base_/models/cgnet.py
new file mode 100644
index 0000000000000000000000000000000000000000..eff8d9458c877c5db894957e0b1b4597e40da6ab
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/cgnet.py
@@ -0,0 +1,35 @@
+# model settings
+norm_cfg = dict(type='SyncBN', eps=1e-03, requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    backbone=dict(
+        type='CGNet',
+        norm_cfg=norm_cfg,
+        in_channels=3,
+        num_channels=(32, 64, 128),
+        num_blocks=(3, 21),
+        dilations=(2, 4),
+        reductions=(8, 16)),
+    decode_head=dict(
+        type='FCNHead',
+        in_channels=256,
+        in_index=2,
+        channels=256,
+        num_convs=0,
+        concat_input=False,
+        dropout_ratio=0,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        loss_decode=dict(
+            type='CrossEntropyLoss',
+            use_sigmoid=False,
+            loss_weight=1.0,
+            class_weight=[
+                2.5959933, 6.7415504, 3.5354059, 9.8663225, 9.690899, 9.369352,
+                10.289121, 9.953208, 4.3097677, 9.490387, 7.674431, 9.396905,
+                10.347791, 6.3927646, 10.226669, 10.241062, 10.280587,
+                10.396974, 10.055647
+            ])),
+    # model training and testing settings
+    train_cfg=dict(sampler=None),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/danet_r50-d8.py b/annotator/uniformer/configs/_base_/models/danet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..2c934939fac48525f22ad86f489a041dd7db7d09
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/danet_r50-d8.py
@@ -0,0 +1,44 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='DAHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        pam_channels=64,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/deeplabv3_r50-d8.py b/annotator/uniformer/configs/_base_/models/deeplabv3_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..d7a43bee01422ad4795dd27874e0cd4bb6cbfecf
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/deeplabv3_r50-d8.py
@@ -0,0 +1,44 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='ASPPHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        dilations=(1, 12, 24, 36),
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/deeplabv3_unet_s5-d16.py b/annotator/uniformer/configs/_base_/models/deeplabv3_unet_s5-d16.py
new file mode 100644
index 0000000000000000000000000000000000000000..0cd262999d8b2cb8e14a5c32190ae73f479d8e81
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/deeplabv3_unet_s5-d16.py
@@ -0,0 +1,50 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained=None,
+    backbone=dict(
+        type='UNet',
+        in_channels=3,
+        base_channels=64,
+        num_stages=5,
+        strides=(1, 1, 1, 1, 1),
+        enc_num_convs=(2, 2, 2, 2, 2),
+        dec_num_convs=(2, 2, 2, 2),
+        downsamples=(True, True, True, True),
+        enc_dilations=(1, 1, 1, 1, 1),
+        dec_dilations=(1, 1, 1, 1),
+        with_cp=False,
+        conv_cfg=None,
+        norm_cfg=norm_cfg,
+        act_cfg=dict(type='ReLU'),
+        upsample_cfg=dict(type='InterpConv'),
+        norm_eval=False),
+    decode_head=dict(
+        type='ASPPHead',
+        in_channels=64,
+        in_index=4,
+        channels=16,
+        dilations=(1, 12, 24, 36),
+        dropout_ratio=0.1,
+        num_classes=2,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=128,
+        in_index=3,
+        channels=64,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=2,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='slide', crop_size=256, stride=170))
diff --git a/annotator/uniformer/configs/_base_/models/deeplabv3plus_r50-d8.py b/annotator/uniformer/configs/_base_/models/deeplabv3plus_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..050e39e091d816df9028d23aa3ecf9db74e441e1
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/deeplabv3plus_r50-d8.py
@@ -0,0 +1,46 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='DepthwiseSeparableASPPHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        dilations=(1, 12, 24, 36),
+        c1_in_channels=256,
+        c1_channels=48,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/dmnet_r50-d8.py b/annotator/uniformer/configs/_base_/models/dmnet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..d22ba52640bebd805b3b8d07025e276dfb023759
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/dmnet_r50-d8.py
@@ -0,0 +1,44 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='DMHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        filter_sizes=(1, 3, 5, 7),
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=dict(type='SyncBN', requires_grad=True),
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/dnl_r50-d8.py b/annotator/uniformer/configs/_base_/models/dnl_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..edb4c174c51e34c103737ba39bfc48bf831e561d
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/dnl_r50-d8.py
@@ -0,0 +1,46 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='DNLHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        dropout_ratio=0.1,
+        reduction=2,
+        use_scale=True,
+        mode='embedded_gaussian',
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/emanet_r50-d8.py b/annotator/uniformer/configs/_base_/models/emanet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..26adcd430926de0862204a71d345f2543167f27b
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/emanet_r50-d8.py
@@ -0,0 +1,47 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='EMAHead',
+        in_channels=2048,
+        in_index=3,
+        channels=256,
+        ema_channels=512,
+        num_bases=64,
+        num_stages=3,
+        momentum=0.1,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/encnet_r50-d8.py b/annotator/uniformer/configs/_base_/models/encnet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..be777123a886503172a95fe0719e956a147bbd68
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/encnet_r50-d8.py
@@ -0,0 +1,48 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='EncHead',
+        in_channels=[512, 1024, 2048],
+        in_index=(1, 2, 3),
+        channels=512,
+        num_codes=32,
+        use_se_loss=True,
+        add_lateral=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
+        loss_se_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=0.2)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/fast_scnn.py b/annotator/uniformer/configs/_base_/models/fast_scnn.py
new file mode 100644
index 0000000000000000000000000000000000000000..32fdeb659355a5ce5ef2cc7c2f30742703811cdf
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/fast_scnn.py
@@ -0,0 +1,57 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True, momentum=0.01)
+model = dict(
+    type='EncoderDecoder',
+    backbone=dict(
+        type='FastSCNN',
+        downsample_dw_channels=(32, 48),
+        global_in_channels=64,
+        global_block_channels=(64, 96, 128),
+        global_block_strides=(2, 2, 1),
+        global_out_channels=128,
+        higher_in_channels=64,
+        lower_in_channels=128,
+        fusion_out_channels=128,
+        out_indices=(0, 1, 2),
+        norm_cfg=norm_cfg,
+        align_corners=False),
+    decode_head=dict(
+        type='DepthwiseSeparableFCNHead',
+        in_channels=128,
+        channels=128,
+        concat_input=False,
+        num_classes=19,
+        in_index=-1,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=0.4)),
+    auxiliary_head=[
+        dict(
+            type='FCNHead',
+            in_channels=128,
+            channels=32,
+            num_convs=1,
+            num_classes=19,
+            in_index=-2,
+            norm_cfg=norm_cfg,
+            concat_input=False,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=0.4)),
+        dict(
+            type='FCNHead',
+            in_channels=64,
+            channels=32,
+            num_convs=1,
+            num_classes=19,
+            in_index=-3,
+            norm_cfg=norm_cfg,
+            concat_input=False,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=True, loss_weight=0.4)),
+    ],
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/fcn_hr18.py b/annotator/uniformer/configs/_base_/models/fcn_hr18.py
new file mode 100644
index 0000000000000000000000000000000000000000..c3e299bc89ada56ca14bbffcbdb08a586b8ed9e9
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/fcn_hr18.py
@@ -0,0 +1,52 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://msra/hrnetv2_w18',
+    backbone=dict(
+        type='HRNet',
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        extra=dict(
+            stage1=dict(
+                num_modules=1,
+                num_branches=1,
+                block='BOTTLENECK',
+                num_blocks=(4, ),
+                num_channels=(64, )),
+            stage2=dict(
+                num_modules=1,
+                num_branches=2,
+                block='BASIC',
+                num_blocks=(4, 4),
+                num_channels=(18, 36)),
+            stage3=dict(
+                num_modules=4,
+                num_branches=3,
+                block='BASIC',
+                num_blocks=(4, 4, 4),
+                num_channels=(18, 36, 72)),
+            stage4=dict(
+                num_modules=3,
+                num_branches=4,
+                block='BASIC',
+                num_blocks=(4, 4, 4, 4),
+                num_channels=(18, 36, 72, 144)))),
+    decode_head=dict(
+        type='FCNHead',
+        in_channels=[18, 36, 72, 144],
+        in_index=(0, 1, 2, 3),
+        channels=sum([18, 36, 72, 144]),
+        input_transform='resize_concat',
+        kernel_size=1,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=-1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/fcn_r50-d8.py b/annotator/uniformer/configs/_base_/models/fcn_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..5e98f6cc918b6146fc6d613c6918e825ef1355c3
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/fcn_r50-d8.py
@@ -0,0 +1,45 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='FCNHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        num_convs=2,
+        concat_input=True,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/fcn_unet_s5-d16.py b/annotator/uniformer/configs/_base_/models/fcn_unet_s5-d16.py
new file mode 100644
index 0000000000000000000000000000000000000000..a33e7972877f902d0e7d18401ca675e3e4e60a18
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/fcn_unet_s5-d16.py
@@ -0,0 +1,51 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained=None,
+    backbone=dict(
+        type='UNet',
+        in_channels=3,
+        base_channels=64,
+        num_stages=5,
+        strides=(1, 1, 1, 1, 1),
+        enc_num_convs=(2, 2, 2, 2, 2),
+        dec_num_convs=(2, 2, 2, 2),
+        downsamples=(True, True, True, True),
+        enc_dilations=(1, 1, 1, 1, 1),
+        dec_dilations=(1, 1, 1, 1),
+        with_cp=False,
+        conv_cfg=None,
+        norm_cfg=norm_cfg,
+        act_cfg=dict(type='ReLU'),
+        upsample_cfg=dict(type='InterpConv'),
+        norm_eval=False),
+    decode_head=dict(
+        type='FCNHead',
+        in_channels=64,
+        in_index=4,
+        channels=64,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=2,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=128,
+        in_index=3,
+        channels=64,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=2,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='slide', crop_size=256, stride=170))
diff --git a/annotator/uniformer/configs/_base_/models/fpn_r50.py b/annotator/uniformer/configs/_base_/models/fpn_r50.py
new file mode 100644
index 0000000000000000000000000000000000000000..86ab327db92e44c14822d65f1c9277cb007f17c1
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/fpn_r50.py
@@ -0,0 +1,36 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 1, 1),
+        strides=(1, 2, 2, 2),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        num_outs=4),
+    decode_head=dict(
+        type='FPNHead',
+        in_channels=[256, 256, 256, 256],
+        in_index=[0, 1, 2, 3],
+        feature_strides=[4, 8, 16, 32],
+        channels=128,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/fpn_uniformer.py b/annotator/uniformer/configs/_base_/models/fpn_uniformer.py
new file mode 100644
index 0000000000000000000000000000000000000000..8aae98c5991055bfcc08e82ccdc09f8b1d9f8a8d
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/fpn_uniformer.py
@@ -0,0 +1,35 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    backbone=dict(
+        type='UniFormer',
+        embed_dim=[64, 128, 320, 512],
+        layers=[3, 4, 8, 3],
+        head_dim=64,
+        mlp_ratio=4.,
+        qkv_bias=True,
+        drop_rate=0.,
+        attn_drop_rate=0.,
+        drop_path_rate=0.1),
+    neck=dict(
+        type='FPN',
+        in_channels=[64, 128, 320, 512],
+        out_channels=256,
+        num_outs=4),
+    decode_head=dict(
+        type='FPNHead',
+        in_channels=[256, 256, 256, 256],
+        in_index=[0, 1, 2, 3],
+        feature_strides=[4, 8, 16, 32],
+        channels=128,
+        dropout_ratio=0.1,
+        num_classes=150,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole')
+)
diff --git a/annotator/uniformer/configs/_base_/models/gcnet_r50-d8.py b/annotator/uniformer/configs/_base_/models/gcnet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..3d2ad69f5c22adfe79d5fdabf920217628987166
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/gcnet_r50-d8.py
@@ -0,0 +1,46 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='GCHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        ratio=1 / 4.,
+        pooling_type='att',
+        fusion_types=('channel_add', ),
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/lraspp_m-v3-d8.py b/annotator/uniformer/configs/_base_/models/lraspp_m-v3-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..93258242a90695cc94a7c6bd41562d6a75988771
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/lraspp_m-v3-d8.py
@@ -0,0 +1,25 @@
+# model settings
+norm_cfg = dict(type='SyncBN', eps=0.001, requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    backbone=dict(
+        type='MobileNetV3',
+        arch='large',
+        out_indices=(1, 3, 16),
+        norm_cfg=norm_cfg),
+    decode_head=dict(
+        type='LRASPPHead',
+        in_channels=(16, 24, 960),
+        in_index=(0, 1, 2),
+        channels=128,
+        input_transform='multiple_select',
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        act_cfg=dict(type='ReLU'),
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/nonlocal_r50-d8.py b/annotator/uniformer/configs/_base_/models/nonlocal_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..5674a39854cafd1f2e363bac99c58ccae62f24da
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/nonlocal_r50-d8.py
@@ -0,0 +1,46 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='NLHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        dropout_ratio=0.1,
+        reduction=2,
+        use_scale=True,
+        mode='embedded_gaussian',
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/ocrnet_hr18.py b/annotator/uniformer/configs/_base_/models/ocrnet_hr18.py
new file mode 100644
index 0000000000000000000000000000000000000000..c60f62a7cdf3f5c5096a7a7e725e8268fddcb057
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/ocrnet_hr18.py
@@ -0,0 +1,68 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='CascadeEncoderDecoder',
+    num_stages=2,
+    pretrained='open-mmlab://msra/hrnetv2_w18',
+    backbone=dict(
+        type='HRNet',
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        extra=dict(
+            stage1=dict(
+                num_modules=1,
+                num_branches=1,
+                block='BOTTLENECK',
+                num_blocks=(4, ),
+                num_channels=(64, )),
+            stage2=dict(
+                num_modules=1,
+                num_branches=2,
+                block='BASIC',
+                num_blocks=(4, 4),
+                num_channels=(18, 36)),
+            stage3=dict(
+                num_modules=4,
+                num_branches=3,
+                block='BASIC',
+                num_blocks=(4, 4, 4),
+                num_channels=(18, 36, 72)),
+            stage4=dict(
+                num_modules=3,
+                num_branches=4,
+                block='BASIC',
+                num_blocks=(4, 4, 4, 4),
+                num_channels=(18, 36, 72, 144)))),
+    decode_head=[
+        dict(
+            type='FCNHead',
+            in_channels=[18, 36, 72, 144],
+            channels=sum([18, 36, 72, 144]),
+            in_index=(0, 1, 2, 3),
+            input_transform='resize_concat',
+            kernel_size=1,
+            num_convs=1,
+            concat_input=False,
+            dropout_ratio=-1,
+            num_classes=19,
+            norm_cfg=norm_cfg,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+        dict(
+            type='OCRHead',
+            in_channels=[18, 36, 72, 144],
+            in_index=(0, 1, 2, 3),
+            input_transform='resize_concat',
+            channels=512,
+            ocr_channels=256,
+            dropout_ratio=-1,
+            num_classes=19,
+            norm_cfg=norm_cfg,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    ],
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/ocrnet_r50-d8.py b/annotator/uniformer/configs/_base_/models/ocrnet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..615aa3ff703942b6c22b2d6e9642504dd3e41ebd
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/ocrnet_r50-d8.py
@@ -0,0 +1,47 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='CascadeEncoderDecoder',
+    num_stages=2,
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=[
+        dict(
+            type='FCNHead',
+            in_channels=1024,
+            in_index=2,
+            channels=256,
+            num_convs=1,
+            concat_input=False,
+            dropout_ratio=0.1,
+            num_classes=19,
+            norm_cfg=norm_cfg,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+        dict(
+            type='OCRHead',
+            in_channels=2048,
+            in_index=3,
+            channels=512,
+            ocr_channels=256,
+            dropout_ratio=0.1,
+            num_classes=19,
+            norm_cfg=norm_cfg,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0))
+    ],
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/pointrend_r50.py b/annotator/uniformer/configs/_base_/models/pointrend_r50.py
new file mode 100644
index 0000000000000000000000000000000000000000..9d323dbf9466d41e0800aa57ef84045f3d874bdf
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/pointrend_r50.py
@@ -0,0 +1,56 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='CascadeEncoderDecoder',
+    num_stages=2,
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 1, 1),
+        strides=(1, 2, 2, 2),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        num_outs=4),
+    decode_head=[
+        dict(
+            type='FPNHead',
+            in_channels=[256, 256, 256, 256],
+            in_index=[0, 1, 2, 3],
+            feature_strides=[4, 8, 16, 32],
+            channels=128,
+            dropout_ratio=-1,
+            num_classes=19,
+            norm_cfg=norm_cfg,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+        dict(
+            type='PointHead',
+            in_channels=[256],
+            in_index=[0],
+            channels=256,
+            num_fcs=3,
+            coarse_pred_each_layer=True,
+            dropout_ratio=-1,
+            num_classes=19,
+            align_corners=False,
+            loss_decode=dict(
+                type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0))
+    ],
+    # model training and testing settings
+    train_cfg=dict(
+        num_points=2048, oversample_ratio=3, importance_sample_ratio=0.75),
+    test_cfg=dict(
+        mode='whole',
+        subdivision_steps=2,
+        subdivision_num_points=8196,
+        scale_factor=2))
diff --git a/annotator/uniformer/configs/_base_/models/psanet_r50-d8.py b/annotator/uniformer/configs/_base_/models/psanet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..689513fa9d2a40f14bf0ae4ae61f38f0dcc1b3da
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/psanet_r50-d8.py
@@ -0,0 +1,49 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='PSAHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        mask_size=(97, 97),
+        psa_type='bi-direction',
+        compact=False,
+        shrink_factor=2,
+        normalization_factor=1.0,
+        psa_softmax=True,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/pspnet_r50-d8.py b/annotator/uniformer/configs/_base_/models/pspnet_r50-d8.py
new file mode 100644
index 0000000000000000000000000000000000000000..f451e08ad2eb0732dcb806b1851eb978d4acf136
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/pspnet_r50-d8.py
@@ -0,0 +1,44 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 2, 4),
+        strides=(1, 2, 1, 1),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='PSPHead',
+        in_channels=2048,
+        in_index=3,
+        channels=512,
+        pool_scales=(1, 2, 3, 6),
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/pspnet_unet_s5-d16.py b/annotator/uniformer/configs/_base_/models/pspnet_unet_s5-d16.py
new file mode 100644
index 0000000000000000000000000000000000000000..fcff9ec4f41fad158344ecd77313dc14564f3682
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/pspnet_unet_s5-d16.py
@@ -0,0 +1,50 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained=None,
+    backbone=dict(
+        type='UNet',
+        in_channels=3,
+        base_channels=64,
+        num_stages=5,
+        strides=(1, 1, 1, 1, 1),
+        enc_num_convs=(2, 2, 2, 2, 2),
+        dec_num_convs=(2, 2, 2, 2),
+        downsamples=(True, True, True, True),
+        enc_dilations=(1, 1, 1, 1, 1),
+        dec_dilations=(1, 1, 1, 1),
+        with_cp=False,
+        conv_cfg=None,
+        norm_cfg=norm_cfg,
+        act_cfg=dict(type='ReLU'),
+        upsample_cfg=dict(type='InterpConv'),
+        norm_eval=False),
+    decode_head=dict(
+        type='PSPHead',
+        in_channels=64,
+        in_index=4,
+        channels=16,
+        pool_scales=(1, 2, 3, 6),
+        dropout_ratio=0.1,
+        num_classes=2,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=128,
+        in_index=3,
+        channels=64,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=2,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='slide', crop_size=256, stride=170))
diff --git a/annotator/uniformer/configs/_base_/models/upernet_r50.py b/annotator/uniformer/configs/_base_/models/upernet_r50.py
new file mode 100644
index 0000000000000000000000000000000000000000..10974962fdd7136031fd06de1700f497d355ceaa
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/upernet_r50.py
@@ -0,0 +1,44 @@
+# model settings
+norm_cfg = dict(type='SyncBN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained='open-mmlab://resnet50_v1c',
+    backbone=dict(
+        type='ResNetV1c',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        dilations=(1, 1, 1, 1),
+        strides=(1, 2, 2, 2),
+        norm_cfg=norm_cfg,
+        norm_eval=False,
+        style='pytorch',
+        contract_dilation=True),
+    decode_head=dict(
+        type='UPerHead',
+        in_channels=[256, 512, 1024, 2048],
+        in_index=[0, 1, 2, 3],
+        pool_scales=(1, 2, 3, 6),
+        channels=512,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=1024,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
diff --git a/annotator/uniformer/configs/_base_/models/upernet_uniformer.py b/annotator/uniformer/configs/_base_/models/upernet_uniformer.py
new file mode 100644
index 0000000000000000000000000000000000000000..41aa4db809dc6e2c508e98051f61807d07477903
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/models/upernet_uniformer.py
@@ -0,0 +1,43 @@
+# model settings
+norm_cfg = dict(type='BN', requires_grad=True)
+model = dict(
+    type='EncoderDecoder',
+    pretrained=None,
+    backbone=dict(
+        type='UniFormer',
+        embed_dim=[64, 128, 320, 512],
+        layers=[3, 4, 8, 3],
+        head_dim=64,
+        mlp_ratio=4.,
+        qkv_bias=True,
+        drop_rate=0.,
+        attn_drop_rate=0.,
+        drop_path_rate=0.1),
+    decode_head=dict(
+        type='UPerHead',
+        in_channels=[64, 128, 320, 512],
+        in_index=[0, 1, 2, 3],
+        pool_scales=(1, 2, 3, 6),
+        channels=512,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
+    auxiliary_head=dict(
+        type='FCNHead',
+        in_channels=320,
+        in_index=2,
+        channels=256,
+        num_convs=1,
+        concat_input=False,
+        dropout_ratio=0.1,
+        num_classes=19,
+        norm_cfg=norm_cfg,
+        align_corners=False,
+        loss_decode=dict(
+            type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
+    # model training and testing settings
+    train_cfg=dict(),
+    test_cfg=dict(mode='whole'))
\ No newline at end of file
diff --git a/annotator/uniformer/configs/_base_/schedules/schedule_160k.py b/annotator/uniformer/configs/_base_/schedules/schedule_160k.py
new file mode 100644
index 0000000000000000000000000000000000000000..52603890b10f25faf8eec9f9e5a4468fae09b811
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/schedules/schedule_160k.py
@@ -0,0 +1,9 @@
+# optimizer
+optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0005)
+optimizer_config = dict()
+# learning policy
+lr_config = dict(policy='poly', power=0.9, min_lr=1e-4, by_epoch=False)
+# runtime settings
+runner = dict(type='IterBasedRunner', max_iters=160000)
+checkpoint_config = dict(by_epoch=False, interval=16000)
+evaluation = dict(interval=16000, metric='mIoU')
diff --git a/annotator/uniformer/configs/_base_/schedules/schedule_20k.py b/annotator/uniformer/configs/_base_/schedules/schedule_20k.py
new file mode 100644
index 0000000000000000000000000000000000000000..bf780a1b6f6521833c6a5859675147824efa599d
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/schedules/schedule_20k.py
@@ -0,0 +1,9 @@
+# optimizer
+optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0005)
+optimizer_config = dict()
+# learning policy
+lr_config = dict(policy='poly', power=0.9, min_lr=1e-4, by_epoch=False)
+# runtime settings
+runner = dict(type='IterBasedRunner', max_iters=20000)
+checkpoint_config = dict(by_epoch=False, interval=2000)
+evaluation = dict(interval=2000, metric='mIoU')
diff --git a/annotator/uniformer/configs/_base_/schedules/schedule_40k.py b/annotator/uniformer/configs/_base_/schedules/schedule_40k.py
new file mode 100644
index 0000000000000000000000000000000000000000..cdbf841abcb26eed87bf76ab816aff4bae0630ee
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/schedules/schedule_40k.py
@@ -0,0 +1,9 @@
+# optimizer
+optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0005)
+optimizer_config = dict()
+# learning policy
+lr_config = dict(policy='poly', power=0.9, min_lr=1e-4, by_epoch=False)
+# runtime settings
+runner = dict(type='IterBasedRunner', max_iters=40000)
+checkpoint_config = dict(by_epoch=False, interval=4000)
+evaluation = dict(interval=4000, metric='mIoU')
diff --git a/annotator/uniformer/configs/_base_/schedules/schedule_80k.py b/annotator/uniformer/configs/_base_/schedules/schedule_80k.py
new file mode 100644
index 0000000000000000000000000000000000000000..c190cee6bdc7922b688ea75dc8f152fa15c24617
--- /dev/null
+++ b/annotator/uniformer/configs/_base_/schedules/schedule_80k.py
@@ -0,0 +1,9 @@
+# optimizer
+optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0005)
+optimizer_config = dict()
+# learning policy
+lr_config = dict(policy='poly', power=0.9, min_lr=1e-4, by_epoch=False)
+# runtime settings
+runner = dict(type='IterBasedRunner', max_iters=80000)
+checkpoint_config = dict(by_epoch=False, interval=8000)
+evaluation = dict(interval=8000, metric='mIoU')
diff --git a/annotator/uniformer/exp/upernet_global_small/config.py b/annotator/uniformer/exp/upernet_global_small/config.py
new file mode 100644
index 0000000000000000000000000000000000000000..01db96bf9b0be531aa0eaf62fee51543712f8670
--- /dev/null
+++ b/annotator/uniformer/exp/upernet_global_small/config.py
@@ -0,0 +1,38 @@
+_base_ = [
+    '../../configs/_base_/models/upernet_uniformer.py', 
+    '../../configs/_base_/datasets/ade20k.py',
+    '../../configs/_base_/default_runtime.py', 
+    '../../configs/_base_/schedules/schedule_160k.py'
+]
+model = dict(
+    backbone=dict(
+        type='UniFormer',
+        embed_dim=[64, 128, 320, 512],
+        layers=[3, 4, 8, 3],
+        head_dim=64,
+        drop_path_rate=0.25,
+        windows=False,
+        hybrid=False
+    ),
+    decode_head=dict(
+        in_channels=[64, 128, 320, 512],
+        num_classes=150
+    ),
+    auxiliary_head=dict(
+        in_channels=320,
+        num_classes=150
+    ))
+
+# AdamW optimizer, no weight decay for position embedding & layer norm in backbone
+optimizer = dict(_delete_=True, type='AdamW', lr=0.00006, betas=(0.9, 0.999), weight_decay=0.01,
+                 paramwise_cfg=dict(custom_keys={'absolute_pos_embed': dict(decay_mult=0.),
+                                                 'relative_position_bias_table': dict(decay_mult=0.),
+                                                 'norm': dict(decay_mult=0.)}))
+
+lr_config = dict(_delete_=True, policy='poly',
+                 warmup='linear',
+                 warmup_iters=1500,
+                 warmup_ratio=1e-6,
+                 power=1.0, min_lr=0.0, by_epoch=False)
+
+data=dict(samples_per_gpu=2)
\ No newline at end of file
diff --git a/annotator/uniformer/exp/upernet_global_small/run.sh b/annotator/uniformer/exp/upernet_global_small/run.sh
new file mode 100644
index 0000000000000000000000000000000000000000..9fb22edfa7a32624ea08a63fe7d720c40db3b696
--- /dev/null
+++ b/annotator/uniformer/exp/upernet_global_small/run.sh
@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+
+work_path=$(dirname $0)
+PYTHONPATH="$(dirname $0)/../../":$PYTHONPATH \
+python -m torch.distributed.launch --nproc_per_node=8 \
+    tools/train.py ${work_path}/config.py \
+    --launcher pytorch \
+    --options model.backbone.pretrained_path='your_model_path/uniformer_small_in1k.pth' \
+    --work-dir ${work_path}/ckpt \
+    2>&1 | tee -a ${work_path}/log.txt
diff --git a/annotator/uniformer/exp/upernet_global_small/test.sh b/annotator/uniformer/exp/upernet_global_small/test.sh
new file mode 100644
index 0000000000000000000000000000000000000000..d9a85e7a0d3b7c96b060f473d41254b37a382fcb
--- /dev/null
+++ b/annotator/uniformer/exp/upernet_global_small/test.sh
@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+
+work_path=$(dirname $0)
+PYTHONPATH="$(dirname $0)/../../":$PYTHONPATH \
+python -m torch.distributed.launch --nproc_per_node=8 \
+    tools/test.py ${work_path}/test_config_h32.py \
+    ${work_path}/ckpt/latest.pth \
+    --launcher pytorch \
+    --eval mIoU \
+    2>&1 | tee -a ${work_path}/log.txt
diff --git a/annotator/uniformer/exp/upernet_global_small/test_config_g.py b/annotator/uniformer/exp/upernet_global_small/test_config_g.py
new file mode 100644
index 0000000000000000000000000000000000000000..e43737a98a3b174a9f2fe059c06d511144686459
--- /dev/null
+++ b/annotator/uniformer/exp/upernet_global_small/test_config_g.py
@@ -0,0 +1,38 @@
+_base_ = [
+    '../../configs/_base_/models/upernet_uniformer.py', 
+    '../../configs/_base_/datasets/ade20k.py',
+    '../../configs/_base_/default_runtime.py', 
+    '../../configs/_base_/schedules/schedule_160k.py'
+]
+model = dict(
+    backbone=dict(
+        type='UniFormer',
+        embed_dim=[64, 128, 320, 512],
+        layers=[3, 4, 8, 3],
+        head_dim=64,
+        drop_path_rate=0.25,
+        windows=False,
+        hybrid=False,
+    ),
+    decode_head=dict(
+        in_channels=[64, 128, 320, 512],
+        num_classes=150
+    ),
+    auxiliary_head=dict(
+        in_channels=320,
+        num_classes=150
+    ))
+
+# AdamW optimizer, no weight decay for position embedding & layer norm in backbone
+optimizer = dict(_delete_=True, type='AdamW', lr=0.00006, betas=(0.9, 0.999), weight_decay=0.01,
+                 paramwise_cfg=dict(custom_keys={'absolute_pos_embed': dict(decay_mult=0.),
+                                                 'relative_position_bias_table': dict(decay_mult=0.),
+                                                 'norm': dict(decay_mult=0.)}))
+
+lr_config = dict(_delete_=True, policy='poly',
+                 warmup='linear',
+                 warmup_iters=1500,
+                 warmup_ratio=1e-6,
+                 power=1.0, min_lr=0.0, by_epoch=False)
+
+data=dict(samples_per_gpu=2)
\ No newline at end of file
diff --git a/annotator/uniformer/exp/upernet_global_small/test_config_h32.py b/annotator/uniformer/exp/upernet_global_small/test_config_h32.py
new file mode 100644
index 0000000000000000000000000000000000000000..a31e3874f76f9f7b089ac8834d85df2441af9b0e
--- /dev/null
+++ b/annotator/uniformer/exp/upernet_global_small/test_config_h32.py
@@ -0,0 +1,39 @@
+_base_ = [
+    '../../configs/_base_/models/upernet_uniformer.py', 
+    '../../configs/_base_/datasets/ade20k.py',
+    '../../configs/_base_/default_runtime.py', 
+    '../../configs/_base_/schedules/schedule_160k.py'
+]
+model = dict(
+    backbone=dict(
+        type='UniFormer',
+        embed_dim=[64, 128, 320, 512],
+        layers=[3, 4, 8, 3],
+        head_dim=64,
+        drop_path_rate=0.25,
+        windows=False,
+        hybrid=True,
+        window_size=32
+    ),
+    decode_head=dict(
+        in_channels=[64, 128, 320, 512],
+        num_classes=150
+    ),
+    auxiliary_head=dict(
+        in_channels=320,
+        num_classes=150
+    ))
+
+# AdamW optimizer, no weight decay for position embedding & layer norm in backbone
+optimizer = dict(_delete_=True, type='AdamW', lr=0.00006, betas=(0.9, 0.999), weight_decay=0.01,
+                 paramwise_cfg=dict(custom_keys={'absolute_pos_embed': dict(decay_mult=0.),
+                                                 'relative_position_bias_table': dict(decay_mult=0.),
+                                                 'norm': dict(decay_mult=0.)}))
+
+lr_config = dict(_delete_=True, policy='poly',
+                 warmup='linear',
+                 warmup_iters=1500,
+                 warmup_ratio=1e-6,
+                 power=1.0, min_lr=0.0, by_epoch=False)
+
+data=dict(samples_per_gpu=2)
\ No newline at end of file
diff --git a/annotator/uniformer/exp/upernet_global_small/test_config_w32.py b/annotator/uniformer/exp/upernet_global_small/test_config_w32.py
new file mode 100644
index 0000000000000000000000000000000000000000..3d9e06f029e46c14cb9ddb39319cabe86fef9b44
--- /dev/null
+++ b/annotator/uniformer/exp/upernet_global_small/test_config_w32.py
@@ -0,0 +1,39 @@
+_base_ = [
+    '../../configs/_base_/models/upernet_uniformer.py', 
+    '../../configs/_base_/datasets/ade20k.py',
+    '../../configs/_base_/default_runtime.py', 
+    '../../configs/_base_/schedules/schedule_160k.py'
+]
+model = dict(
+    backbone=dict(
+        type='UniFormer',
+        embed_dim=[64, 128, 320, 512],
+        layers=[3, 4, 8, 3],
+        head_dim=64,
+        drop_path_rate=0.25,
+        windows=True,
+        hybrid=False,
+        window_size=32
+    ),
+    decode_head=dict(
+        in_channels=[64, 128, 320, 512],
+        num_classes=150
+    ),
+    auxiliary_head=dict(
+        in_channels=320,
+        num_classes=150
+    ))
+
+# AdamW optimizer, no weight decay for position embedding & layer norm in backbone
+optimizer = dict(_delete_=True, type='AdamW', lr=0.00006, betas=(0.9, 0.999), weight_decay=0.01,
+                 paramwise_cfg=dict(custom_keys={'absolute_pos_embed': dict(decay_mult=0.),
+                                                 'relative_position_bias_table': dict(decay_mult=0.),
+                                                 'norm': dict(decay_mult=0.)}))
+
+lr_config = dict(_delete_=True, policy='poly',
+                 warmup='linear',
+                 warmup_iters=1500,
+                 warmup_ratio=1e-6,
+                 power=1.0, min_lr=0.0, by_epoch=False)
+
+data=dict(samples_per_gpu=2)
\ No newline at end of file
diff --git a/annotator/uniformer/mmcv/__init__.py b/annotator/uniformer/mmcv/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..210a2989138380559f23045b568d0fbbeb918c03
--- /dev/null
+++ b/annotator/uniformer/mmcv/__init__.py
@@ -0,0 +1,15 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+# flake8: noqa
+from .arraymisc import *
+from .fileio import *
+from .image import *
+from .utils import *
+from .version import *
+from .video import *
+from .visualization import *
+
+# The following modules are not imported to this level, so mmcv may be used
+# without PyTorch.
+# - runner
+# - parallel
+# - op
diff --git a/annotator/uniformer/mmcv/arraymisc/__init__.py b/annotator/uniformer/mmcv/arraymisc/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..4b4700d6139ae3d604ff6e542468cce4200c020c
--- /dev/null
+++ b/annotator/uniformer/mmcv/arraymisc/__init__.py
@@ -0,0 +1,4 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .quantization import dequantize, quantize
+
+__all__ = ['quantize', 'dequantize']
diff --git a/annotator/uniformer/mmcv/arraymisc/quantization.py b/annotator/uniformer/mmcv/arraymisc/quantization.py
new file mode 100644
index 0000000000000000000000000000000000000000..8e47a3545780cf071a1ef8195efb0b7b662c8186
--- /dev/null
+++ b/annotator/uniformer/mmcv/arraymisc/quantization.py
@@ -0,0 +1,55 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import numpy as np
+
+
+def quantize(arr, min_val, max_val, levels, dtype=np.int64):
+    """Quantize an array of (-inf, inf) to [0, levels-1].
+
+    Args:
+        arr (ndarray): Input array.
+        min_val (scalar): Minimum value to be clipped.
+        max_val (scalar): Maximum value to be clipped.
+        levels (int): Quantization levels.
+        dtype (np.type): The type of the quantized array.
+
+    Returns:
+        tuple: Quantized array.
+    """
+    if not (isinstance(levels, int) and levels > 1):
+        raise ValueError(
+            f'levels must be a positive integer, but got {levels}')
+    if min_val >= max_val:
+        raise ValueError(
+            f'min_val ({min_val}) must be smaller than max_val ({max_val})')
+
+    arr = np.clip(arr, min_val, max_val) - min_val
+    quantized_arr = np.minimum(
+        np.floor(levels * arr / (max_val - min_val)).astype(dtype), levels - 1)
+
+    return quantized_arr
+
+
+def dequantize(arr, min_val, max_val, levels, dtype=np.float64):
+    """Dequantize an array.
+
+    Args:
+        arr (ndarray): Input array.
+        min_val (scalar): Minimum value to be clipped.
+        max_val (scalar): Maximum value to be clipped.
+        levels (int): Quantization levels.
+        dtype (np.type): The type of the dequantized array.
+
+    Returns:
+        tuple: Dequantized array.
+    """
+    if not (isinstance(levels, int) and levels > 1):
+        raise ValueError(
+            f'levels must be a positive integer, but got {levels}')
+    if min_val >= max_val:
+        raise ValueError(
+            f'min_val ({min_val}) must be smaller than max_val ({max_val})')
+
+    dequantized_arr = (arr + 0.5).astype(dtype) * (max_val -
+                                                   min_val) / levels + min_val
+
+    return dequantized_arr
diff --git a/annotator/uniformer/mmcv/cnn/__init__.py b/annotator/uniformer/mmcv/cnn/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..7246c897430f0cc7ce12719ad8608824fc734446
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/__init__.py
@@ -0,0 +1,41 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .alexnet import AlexNet
+# yapf: disable
+from .bricks import (ACTIVATION_LAYERS, CONV_LAYERS, NORM_LAYERS,
+                     PADDING_LAYERS, PLUGIN_LAYERS, UPSAMPLE_LAYERS,
+                     ContextBlock, Conv2d, Conv3d, ConvAWS2d, ConvModule,
+                     ConvTranspose2d, ConvTranspose3d, ConvWS2d,
+                     DepthwiseSeparableConvModule, GeneralizedAttention,
+                     HSigmoid, HSwish, Linear, MaxPool2d, MaxPool3d,
+                     NonLocal1d, NonLocal2d, NonLocal3d, Scale, Swish,
+                     build_activation_layer, build_conv_layer,
+                     build_norm_layer, build_padding_layer, build_plugin_layer,
+                     build_upsample_layer, conv_ws_2d, is_norm)
+from .builder import MODELS, build_model_from_cfg
+# yapf: enable
+from .resnet import ResNet, make_res_layer
+from .utils import (INITIALIZERS, Caffe2XavierInit, ConstantInit, KaimingInit,
+                    NormalInit, PretrainedInit, TruncNormalInit, UniformInit,
+                    XavierInit, bias_init_with_prob, caffe2_xavier_init,
+                    constant_init, fuse_conv_bn, get_model_complexity_info,
+                    initialize, kaiming_init, normal_init, trunc_normal_init,
+                    uniform_init, xavier_init)
+from .vgg import VGG, make_vgg_layer
+
+__all__ = [
+    'AlexNet', 'VGG', 'make_vgg_layer', 'ResNet', 'make_res_layer',
+    'constant_init', 'xavier_init', 'normal_init', 'trunc_normal_init',
+    'uniform_init', 'kaiming_init', 'caffe2_xavier_init',
+    'bias_init_with_prob', 'ConvModule', 'build_activation_layer',
+    'build_conv_layer', 'build_norm_layer', 'build_padding_layer',
+    'build_upsample_layer', 'build_plugin_layer', 'is_norm', 'NonLocal1d',
+    'NonLocal2d', 'NonLocal3d', 'ContextBlock', 'HSigmoid', 'Swish', 'HSwish',
+    'GeneralizedAttention', 'ACTIVATION_LAYERS', 'CONV_LAYERS', 'NORM_LAYERS',
+    'PADDING_LAYERS', 'UPSAMPLE_LAYERS', 'PLUGIN_LAYERS', 'Scale',
+    'get_model_complexity_info', 'conv_ws_2d', 'ConvAWS2d', 'ConvWS2d',
+    'fuse_conv_bn', 'DepthwiseSeparableConvModule', 'Linear', 'Conv2d',
+    'ConvTranspose2d', 'MaxPool2d', 'ConvTranspose3d', 'MaxPool3d', 'Conv3d',
+    'initialize', 'INITIALIZERS', 'ConstantInit', 'XavierInit', 'NormalInit',
+    'TruncNormalInit', 'UniformInit', 'KaimingInit', 'PretrainedInit',
+    'Caffe2XavierInit', 'MODELS', 'build_model_from_cfg'
+]
diff --git a/annotator/uniformer/mmcv/cnn/alexnet.py b/annotator/uniformer/mmcv/cnn/alexnet.py
new file mode 100644
index 0000000000000000000000000000000000000000..89e36b8c7851f895d9ae7f07149f0e707456aab0
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/alexnet.py
@@ -0,0 +1,61 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import logging
+
+import torch.nn as nn
+
+
+class AlexNet(nn.Module):
+    """AlexNet backbone.
+
+    Args:
+        num_classes (int): number of classes for classification.
+    """
+
+    def __init__(self, num_classes=-1):
+        super(AlexNet, self).__init__()
+        self.num_classes = num_classes
+        self.features = nn.Sequential(
+            nn.Conv2d(3, 64, kernel_size=11, stride=4, padding=2),
+            nn.ReLU(inplace=True),
+            nn.MaxPool2d(kernel_size=3, stride=2),
+            nn.Conv2d(64, 192, kernel_size=5, padding=2),
+            nn.ReLU(inplace=True),
+            nn.MaxPool2d(kernel_size=3, stride=2),
+            nn.Conv2d(192, 384, kernel_size=3, padding=1),
+            nn.ReLU(inplace=True),
+            nn.Conv2d(384, 256, kernel_size=3, padding=1),
+            nn.ReLU(inplace=True),
+            nn.Conv2d(256, 256, kernel_size=3, padding=1),
+            nn.ReLU(inplace=True),
+            nn.MaxPool2d(kernel_size=3, stride=2),
+        )
+        if self.num_classes > 0:
+            self.classifier = nn.Sequential(
+                nn.Dropout(),
+                nn.Linear(256 * 6 * 6, 4096),
+                nn.ReLU(inplace=True),
+                nn.Dropout(),
+                nn.Linear(4096, 4096),
+                nn.ReLU(inplace=True),
+                nn.Linear(4096, num_classes),
+            )
+
+    def init_weights(self, pretrained=None):
+        if isinstance(pretrained, str):
+            logger = logging.getLogger()
+            from ..runner import load_checkpoint
+            load_checkpoint(self, pretrained, strict=False, logger=logger)
+        elif pretrained is None:
+            # use default initializer
+            pass
+        else:
+            raise TypeError('pretrained must be a str or None')
+
+    def forward(self, x):
+
+        x = self.features(x)
+        if self.num_classes > 0:
+            x = x.view(x.size(0), 256 * 6 * 6)
+            x = self.classifier(x)
+
+        return x
diff --git a/annotator/uniformer/mmcv/cnn/bricks/__init__.py b/annotator/uniformer/mmcv/cnn/bricks/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..0f33124ed23fc6f27119a37bcb5ab004d3572be0
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/__init__.py
@@ -0,0 +1,35 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .activation import build_activation_layer
+from .context_block import ContextBlock
+from .conv import build_conv_layer
+from .conv2d_adaptive_padding import Conv2dAdaptivePadding
+from .conv_module import ConvModule
+from .conv_ws import ConvAWS2d, ConvWS2d, conv_ws_2d
+from .depthwise_separable_conv_module import DepthwiseSeparableConvModule
+from .drop import Dropout, DropPath
+from .generalized_attention import GeneralizedAttention
+from .hsigmoid import HSigmoid
+from .hswish import HSwish
+from .non_local import NonLocal1d, NonLocal2d, NonLocal3d
+from .norm import build_norm_layer, is_norm
+from .padding import build_padding_layer
+from .plugin import build_plugin_layer
+from .registry import (ACTIVATION_LAYERS, CONV_LAYERS, NORM_LAYERS,
+                       PADDING_LAYERS, PLUGIN_LAYERS, UPSAMPLE_LAYERS)
+from .scale import Scale
+from .swish import Swish
+from .upsample import build_upsample_layer
+from .wrappers import (Conv2d, Conv3d, ConvTranspose2d, ConvTranspose3d,
+                       Linear, MaxPool2d, MaxPool3d)
+
+__all__ = [
+    'ConvModule', 'build_activation_layer', 'build_conv_layer',
+    'build_norm_layer', 'build_padding_layer', 'build_upsample_layer',
+    'build_plugin_layer', 'is_norm', 'HSigmoid', 'HSwish', 'NonLocal1d',
+    'NonLocal2d', 'NonLocal3d', 'ContextBlock', 'GeneralizedAttention',
+    'ACTIVATION_LAYERS', 'CONV_LAYERS', 'NORM_LAYERS', 'PADDING_LAYERS',
+    'UPSAMPLE_LAYERS', 'PLUGIN_LAYERS', 'Scale', 'ConvAWS2d', 'ConvWS2d',
+    'conv_ws_2d', 'DepthwiseSeparableConvModule', 'Swish', 'Linear',
+    'Conv2dAdaptivePadding', 'Conv2d', 'ConvTranspose2d', 'MaxPool2d',
+    'ConvTranspose3d', 'MaxPool3d', 'Conv3d', 'Dropout', 'DropPath'
+]
diff --git a/annotator/uniformer/mmcv/cnn/bricks/activation.py b/annotator/uniformer/mmcv/cnn/bricks/activation.py
new file mode 100644
index 0000000000000000000000000000000000000000..cab2712287d5ef7be2f079dcb54a94b96394eab5
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/activation.py
@@ -0,0 +1,92 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from annotator.uniformer.mmcv.utils import TORCH_VERSION, build_from_cfg, digit_version
+from .registry import ACTIVATION_LAYERS
+
+for module in [
+        nn.ReLU, nn.LeakyReLU, nn.PReLU, nn.RReLU, nn.ReLU6, nn.ELU,
+        nn.Sigmoid, nn.Tanh
+]:
+    ACTIVATION_LAYERS.register_module(module=module)
+
+
+@ACTIVATION_LAYERS.register_module(name='Clip')
+@ACTIVATION_LAYERS.register_module()
+class Clamp(nn.Module):
+    """Clamp activation layer.
+
+    This activation function is to clamp the feature map value within
+    :math:`[min, max]`. More details can be found in ``torch.clamp()``.
+
+    Args:
+        min (Number | optional): Lower-bound of the range to be clamped to.
+            Default to -1.
+        max (Number | optional): Upper-bound of the range to be clamped to.
+            Default to 1.
+    """
+
+    def __init__(self, min=-1., max=1.):
+        super(Clamp, self).__init__()
+        self.min = min
+        self.max = max
+
+    def forward(self, x):
+        """Forward function.
+
+        Args:
+            x (torch.Tensor): The input tensor.
+
+        Returns:
+            torch.Tensor: Clamped tensor.
+        """
+        return torch.clamp(x, min=self.min, max=self.max)
+
+
+class GELU(nn.Module):
+    r"""Applies the Gaussian Error Linear Units function:
+
+    .. math::
+        \text{GELU}(x) = x * \Phi(x)
+    where :math:`\Phi(x)` is the Cumulative Distribution Function for
+    Gaussian Distribution.
+
+    Shape:
+        - Input: :math:`(N, *)` where `*` means, any number of additional
+          dimensions
+        - Output: :math:`(N, *)`, same shape as the input
+
+    .. image:: scripts/activation_images/GELU.png
+
+    Examples::
+
+        >>> m = nn.GELU()
+        >>> input = torch.randn(2)
+        >>> output = m(input)
+    """
+
+    def forward(self, input):
+        return F.gelu(input)
+
+
+if (TORCH_VERSION == 'parrots'
+        or digit_version(TORCH_VERSION) < digit_version('1.4')):
+    ACTIVATION_LAYERS.register_module(module=GELU)
+else:
+    ACTIVATION_LAYERS.register_module(module=nn.GELU)
+
+
+def build_activation_layer(cfg):
+    """Build activation layer.
+
+    Args:
+        cfg (dict): The activation layer config, which should contain:
+            - type (str): Layer type.
+            - layer args: Args needed to instantiate an activation layer.
+
+    Returns:
+        nn.Module: Created activation layer.
+    """
+    return build_from_cfg(cfg, ACTIVATION_LAYERS)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/context_block.py b/annotator/uniformer/mmcv/cnn/bricks/context_block.py
new file mode 100644
index 0000000000000000000000000000000000000000..d60fdb904c749ce3b251510dff3cc63cea70d42e
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/context_block.py
@@ -0,0 +1,125 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch import nn
+
+from ..utils import constant_init, kaiming_init
+from .registry import PLUGIN_LAYERS
+
+
+def last_zero_init(m):
+    if isinstance(m, nn.Sequential):
+        constant_init(m[-1], val=0)
+    else:
+        constant_init(m, val=0)
+
+
+@PLUGIN_LAYERS.register_module()
+class ContextBlock(nn.Module):
+    """ContextBlock module in GCNet.
+
+    See 'GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond'
+    (https://arxiv.org/abs/1904.11492) for details.
+
+    Args:
+        in_channels (int): Channels of the input feature map.
+        ratio (float): Ratio of channels of transform bottleneck
+        pooling_type (str): Pooling method for context modeling.
+            Options are 'att' and 'avg', stand for attention pooling and
+            average pooling respectively. Default: 'att'.
+        fusion_types (Sequence[str]): Fusion method for feature fusion,
+            Options are 'channels_add', 'channel_mul', stand for channelwise
+            addition and multiplication respectively. Default: ('channel_add',)
+    """
+
+    _abbr_ = 'context_block'
+
+    def __init__(self,
+                 in_channels,
+                 ratio,
+                 pooling_type='att',
+                 fusion_types=('channel_add', )):
+        super(ContextBlock, self).__init__()
+        assert pooling_type in ['avg', 'att']
+        assert isinstance(fusion_types, (list, tuple))
+        valid_fusion_types = ['channel_add', 'channel_mul']
+        assert all([f in valid_fusion_types for f in fusion_types])
+        assert len(fusion_types) > 0, 'at least one fusion should be used'
+        self.in_channels = in_channels
+        self.ratio = ratio
+        self.planes = int(in_channels * ratio)
+        self.pooling_type = pooling_type
+        self.fusion_types = fusion_types
+        if pooling_type == 'att':
+            self.conv_mask = nn.Conv2d(in_channels, 1, kernel_size=1)
+            self.softmax = nn.Softmax(dim=2)
+        else:
+            self.avg_pool = nn.AdaptiveAvgPool2d(1)
+        if 'channel_add' in fusion_types:
+            self.channel_add_conv = nn.Sequential(
+                nn.Conv2d(self.in_channels, self.planes, kernel_size=1),
+                nn.LayerNorm([self.planes, 1, 1]),
+                nn.ReLU(inplace=True),  # yapf: disable
+                nn.Conv2d(self.planes, self.in_channels, kernel_size=1))
+        else:
+            self.channel_add_conv = None
+        if 'channel_mul' in fusion_types:
+            self.channel_mul_conv = nn.Sequential(
+                nn.Conv2d(self.in_channels, self.planes, kernel_size=1),
+                nn.LayerNorm([self.planes, 1, 1]),
+                nn.ReLU(inplace=True),  # yapf: disable
+                nn.Conv2d(self.planes, self.in_channels, kernel_size=1))
+        else:
+            self.channel_mul_conv = None
+        self.reset_parameters()
+
+    def reset_parameters(self):
+        if self.pooling_type == 'att':
+            kaiming_init(self.conv_mask, mode='fan_in')
+            self.conv_mask.inited = True
+
+        if self.channel_add_conv is not None:
+            last_zero_init(self.channel_add_conv)
+        if self.channel_mul_conv is not None:
+            last_zero_init(self.channel_mul_conv)
+
+    def spatial_pool(self, x):
+        batch, channel, height, width = x.size()
+        if self.pooling_type == 'att':
+            input_x = x
+            # [N, C, H * W]
+            input_x = input_x.view(batch, channel, height * width)
+            # [N, 1, C, H * W]
+            input_x = input_x.unsqueeze(1)
+            # [N, 1, H, W]
+            context_mask = self.conv_mask(x)
+            # [N, 1, H * W]
+            context_mask = context_mask.view(batch, 1, height * width)
+            # [N, 1, H * W]
+            context_mask = self.softmax(context_mask)
+            # [N, 1, H * W, 1]
+            context_mask = context_mask.unsqueeze(-1)
+            # [N, 1, C, 1]
+            context = torch.matmul(input_x, context_mask)
+            # [N, C, 1, 1]
+            context = context.view(batch, channel, 1, 1)
+        else:
+            # [N, C, 1, 1]
+            context = self.avg_pool(x)
+
+        return context
+
+    def forward(self, x):
+        # [N, C, 1, 1]
+        context = self.spatial_pool(x)
+
+        out = x
+        if self.channel_mul_conv is not None:
+            # [N, C, 1, 1]
+            channel_mul_term = torch.sigmoid(self.channel_mul_conv(context))
+            out = out * channel_mul_term
+        if self.channel_add_conv is not None:
+            # [N, C, 1, 1]
+            channel_add_term = self.channel_add_conv(context)
+            out = out + channel_add_term
+
+        return out
diff --git a/annotator/uniformer/mmcv/cnn/bricks/conv.py b/annotator/uniformer/mmcv/cnn/bricks/conv.py
new file mode 100644
index 0000000000000000000000000000000000000000..cf54491997a48ac3e7fadc4183ab7bf3e831024c
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/conv.py
@@ -0,0 +1,44 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from torch import nn
+
+from .registry import CONV_LAYERS
+
+CONV_LAYERS.register_module('Conv1d', module=nn.Conv1d)
+CONV_LAYERS.register_module('Conv2d', module=nn.Conv2d)
+CONV_LAYERS.register_module('Conv3d', module=nn.Conv3d)
+CONV_LAYERS.register_module('Conv', module=nn.Conv2d)
+
+
+def build_conv_layer(cfg, *args, **kwargs):
+    """Build convolution layer.
+
+    Args:
+        cfg (None or dict): The conv layer config, which should contain:
+            - type (str): Layer type.
+            - layer args: Args needed to instantiate an conv layer.
+        args (argument list): Arguments passed to the `__init__`
+            method of the corresponding conv layer.
+        kwargs (keyword arguments): Keyword arguments passed to the `__init__`
+            method of the corresponding conv layer.
+
+    Returns:
+        nn.Module: Created conv layer.
+    """
+    if cfg is None:
+        cfg_ = dict(type='Conv2d')
+    else:
+        if not isinstance(cfg, dict):
+            raise TypeError('cfg must be a dict')
+        if 'type' not in cfg:
+            raise KeyError('the cfg dict must contain the key "type"')
+        cfg_ = cfg.copy()
+
+    layer_type = cfg_.pop('type')
+    if layer_type not in CONV_LAYERS:
+        raise KeyError(f'Unrecognized norm type {layer_type}')
+    else:
+        conv_layer = CONV_LAYERS.get(layer_type)
+
+    layer = conv_layer(*args, **kwargs, **cfg_)
+
+    return layer
diff --git a/annotator/uniformer/mmcv/cnn/bricks/conv2d_adaptive_padding.py b/annotator/uniformer/mmcv/cnn/bricks/conv2d_adaptive_padding.py
new file mode 100644
index 0000000000000000000000000000000000000000..b45e758ac6cf8dfb0382d072fe09125bc7e9b888
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/conv2d_adaptive_padding.py
@@ -0,0 +1,62 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import math
+
+from torch import nn
+from torch.nn import functional as F
+
+from .registry import CONV_LAYERS
+
+
+@CONV_LAYERS.register_module()
+class Conv2dAdaptivePadding(nn.Conv2d):
+    """Implementation of 2D convolution in tensorflow with `padding` as "same",
+    which applies padding to input (if needed) so that input image gets fully
+    covered by filter and stride you specified. For stride 1, this will ensure
+    that output image size is same as input. For stride of 2, output dimensions
+    will be half, for example.
+
+    Args:
+        in_channels (int): Number of channels in the input image
+        out_channels (int): Number of channels produced by the convolution
+        kernel_size (int or tuple): Size of the convolving kernel
+        stride (int or tuple, optional): Stride of the convolution. Default: 1
+        padding (int or tuple, optional): Zero-padding added to both sides of
+            the input. Default: 0
+        dilation (int or tuple, optional): Spacing between kernel elements.
+            Default: 1
+        groups (int, optional): Number of blocked connections from input
+            channels to output channels. Default: 1
+        bias (bool, optional): If ``True``, adds a learnable bias to the
+            output. Default: ``True``
+    """
+
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 groups=1,
+                 bias=True):
+        super().__init__(in_channels, out_channels, kernel_size, stride, 0,
+                         dilation, groups, bias)
+
+    def forward(self, x):
+        img_h, img_w = x.size()[-2:]
+        kernel_h, kernel_w = self.weight.size()[-2:]
+        stride_h, stride_w = self.stride
+        output_h = math.ceil(img_h / stride_h)
+        output_w = math.ceil(img_w / stride_w)
+        pad_h = (
+            max((output_h - 1) * self.stride[0] +
+                (kernel_h - 1) * self.dilation[0] + 1 - img_h, 0))
+        pad_w = (
+            max((output_w - 1) * self.stride[1] +
+                (kernel_w - 1) * self.dilation[1] + 1 - img_w, 0))
+        if pad_h > 0 or pad_w > 0:
+            x = F.pad(x, [
+                pad_w // 2, pad_w - pad_w // 2, pad_h // 2, pad_h - pad_h // 2
+            ])
+        return F.conv2d(x, self.weight, self.bias, self.stride, self.padding,
+                        self.dilation, self.groups)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/conv_module.py b/annotator/uniformer/mmcv/cnn/bricks/conv_module.py
new file mode 100644
index 0000000000000000000000000000000000000000..e60e7e62245071c77b652093fddebff3948d7c3e
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/conv_module.py
@@ -0,0 +1,206 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import warnings
+
+import torch.nn as nn
+
+from annotator.uniformer.mmcv.utils import _BatchNorm, _InstanceNorm
+from ..utils import constant_init, kaiming_init
+from .activation import build_activation_layer
+from .conv import build_conv_layer
+from .norm import build_norm_layer
+from .padding import build_padding_layer
+from .registry import PLUGIN_LAYERS
+
+
+@PLUGIN_LAYERS.register_module()
+class ConvModule(nn.Module):
+    """A conv block that bundles conv/norm/activation layers.
+
+    This block simplifies the usage of convolution layers, which are commonly
+    used with a norm layer (e.g., BatchNorm) and activation layer (e.g., ReLU).
+    It is based upon three build methods: `build_conv_layer()`,
+    `build_norm_layer()` and `build_activation_layer()`.
+
+    Besides, we add some additional features in this module.
+    1. Automatically set `bias` of the conv layer.
+    2. Spectral norm is supported.
+    3. More padding modes are supported. Before PyTorch 1.5, nn.Conv2d only
+    supports zero and circular padding, and we add "reflect" padding mode.
+
+    Args:
+        in_channels (int): Number of channels in the input feature map.
+            Same as that in ``nn._ConvNd``.
+        out_channels (int): Number of channels produced by the convolution.
+            Same as that in ``nn._ConvNd``.
+        kernel_size (int | tuple[int]): Size of the convolving kernel.
+            Same as that in ``nn._ConvNd``.
+        stride (int | tuple[int]): Stride of the convolution.
+            Same as that in ``nn._ConvNd``.
+        padding (int | tuple[int]): Zero-padding added to both sides of
+            the input. Same as that in ``nn._ConvNd``.
+        dilation (int | tuple[int]): Spacing between kernel elements.
+            Same as that in ``nn._ConvNd``.
+        groups (int): Number of blocked connections from input channels to
+            output channels. Same as that in ``nn._ConvNd``.
+        bias (bool | str): If specified as `auto`, it will be decided by the
+            norm_cfg. Bias will be set as True if `norm_cfg` is None, otherwise
+            False. Default: "auto".
+        conv_cfg (dict): Config dict for convolution layer. Default: None,
+            which means using conv2d.
+        norm_cfg (dict): Config dict for normalization layer. Default: None.
+        act_cfg (dict): Config dict for activation layer.
+            Default: dict(type='ReLU').
+        inplace (bool): Whether to use inplace mode for activation.
+            Default: True.
+        with_spectral_norm (bool): Whether use spectral norm in conv module.
+            Default: False.
+        padding_mode (str): If the `padding_mode` has not been supported by
+            current `Conv2d` in PyTorch, we will use our own padding layer
+            instead. Currently, we support ['zeros', 'circular'] with official
+            implementation and ['reflect'] with our own implementation.
+            Default: 'zeros'.
+        order (tuple[str]): The order of conv/norm/activation layers. It is a
+            sequence of "conv", "norm" and "act". Common examples are
+            ("conv", "norm", "act") and ("act", "conv", "norm").
+            Default: ('conv', 'norm', 'act').
+    """
+
+    _abbr_ = 'conv_block'
+
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 groups=1,
+                 bias='auto',
+                 conv_cfg=None,
+                 norm_cfg=None,
+                 act_cfg=dict(type='ReLU'),
+                 inplace=True,
+                 with_spectral_norm=False,
+                 padding_mode='zeros',
+                 order=('conv', 'norm', 'act')):
+        super(ConvModule, self).__init__()
+        assert conv_cfg is None or isinstance(conv_cfg, dict)
+        assert norm_cfg is None or isinstance(norm_cfg, dict)
+        assert act_cfg is None or isinstance(act_cfg, dict)
+        official_padding_mode = ['zeros', 'circular']
+        self.conv_cfg = conv_cfg
+        self.norm_cfg = norm_cfg
+        self.act_cfg = act_cfg
+        self.inplace = inplace
+        self.with_spectral_norm = with_spectral_norm
+        self.with_explicit_padding = padding_mode not in official_padding_mode
+        self.order = order
+        assert isinstance(self.order, tuple) and len(self.order) == 3
+        assert set(order) == set(['conv', 'norm', 'act'])
+
+        self.with_norm = norm_cfg is not None
+        self.with_activation = act_cfg is not None
+        # if the conv layer is before a norm layer, bias is unnecessary.
+        if bias == 'auto':
+            bias = not self.with_norm
+        self.with_bias = bias
+
+        if self.with_explicit_padding:
+            pad_cfg = dict(type=padding_mode)
+            self.padding_layer = build_padding_layer(pad_cfg, padding)
+
+        # reset padding to 0 for conv module
+        conv_padding = 0 if self.with_explicit_padding else padding
+        # build convolution layer
+        self.conv = build_conv_layer(
+            conv_cfg,
+            in_channels,
+            out_channels,
+            kernel_size,
+            stride=stride,
+            padding=conv_padding,
+            dilation=dilation,
+            groups=groups,
+            bias=bias)
+        # export the attributes of self.conv to a higher level for convenience
+        self.in_channels = self.conv.in_channels
+        self.out_channels = self.conv.out_channels
+        self.kernel_size = self.conv.kernel_size
+        self.stride = self.conv.stride
+        self.padding = padding
+        self.dilation = self.conv.dilation
+        self.transposed = self.conv.transposed
+        self.output_padding = self.conv.output_padding
+        self.groups = self.conv.groups
+
+        if self.with_spectral_norm:
+            self.conv = nn.utils.spectral_norm(self.conv)
+
+        # build normalization layers
+        if self.with_norm:
+            # norm layer is after conv layer
+            if order.index('norm') > order.index('conv'):
+                norm_channels = out_channels
+            else:
+                norm_channels = in_channels
+            self.norm_name, norm = build_norm_layer(norm_cfg, norm_channels)
+            self.add_module(self.norm_name, norm)
+            if self.with_bias:
+                if isinstance(norm, (_BatchNorm, _InstanceNorm)):
+                    warnings.warn(
+                        'Unnecessary conv bias before batch/instance norm')
+        else:
+            self.norm_name = None
+
+        # build activation layer
+        if self.with_activation:
+            act_cfg_ = act_cfg.copy()
+            # nn.Tanh has no 'inplace' argument
+            if act_cfg_['type'] not in [
+                    'Tanh', 'PReLU', 'Sigmoid', 'HSigmoid', 'Swish'
+            ]:
+                act_cfg_.setdefault('inplace', inplace)
+            self.activate = build_activation_layer(act_cfg_)
+
+        # Use msra init by default
+        self.init_weights()
+
+    @property
+    def norm(self):
+        if self.norm_name:
+            return getattr(self, self.norm_name)
+        else:
+            return None
+
+    def init_weights(self):
+        # 1. It is mainly for customized conv layers with their own
+        #    initialization manners by calling their own ``init_weights()``,
+        #    and we do not want ConvModule to override the initialization.
+        # 2. For customized conv layers without their own initialization
+        #    manners (that is, they don't have their own ``init_weights()``)
+        #    and PyTorch's conv layers, they will be initialized by
+        #    this method with default ``kaiming_init``.
+        # Note: For PyTorch's conv layers, they will be overwritten by our
+        #    initialization implementation using default ``kaiming_init``.
+        if not hasattr(self.conv, 'init_weights'):
+            if self.with_activation and self.act_cfg['type'] == 'LeakyReLU':
+                nonlinearity = 'leaky_relu'
+                a = self.act_cfg.get('negative_slope', 0.01)
+            else:
+                nonlinearity = 'relu'
+                a = 0
+            kaiming_init(self.conv, a=a, nonlinearity=nonlinearity)
+        if self.with_norm:
+            constant_init(self.norm, 1, bias=0)
+
+    def forward(self, x, activate=True, norm=True):
+        for layer in self.order:
+            if layer == 'conv':
+                if self.with_explicit_padding:
+                    x = self.padding_layer(x)
+                x = self.conv(x)
+            elif layer == 'norm' and norm and self.with_norm:
+                x = self.norm(x)
+            elif layer == 'act' and activate and self.with_activation:
+                x = self.activate(x)
+        return x
diff --git a/annotator/uniformer/mmcv/cnn/bricks/conv_ws.py b/annotator/uniformer/mmcv/cnn/bricks/conv_ws.py
new file mode 100644
index 0000000000000000000000000000000000000000..a3941e27874993418b3b5708d5a7485f175ff9c8
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/conv_ws.py
@@ -0,0 +1,148 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from .registry import CONV_LAYERS
+
+
+def conv_ws_2d(input,
+               weight,
+               bias=None,
+               stride=1,
+               padding=0,
+               dilation=1,
+               groups=1,
+               eps=1e-5):
+    c_in = weight.size(0)
+    weight_flat = weight.view(c_in, -1)
+    mean = weight_flat.mean(dim=1, keepdim=True).view(c_in, 1, 1, 1)
+    std = weight_flat.std(dim=1, keepdim=True).view(c_in, 1, 1, 1)
+    weight = (weight - mean) / (std + eps)
+    return F.conv2d(input, weight, bias, stride, padding, dilation, groups)
+
+
+@CONV_LAYERS.register_module('ConvWS')
+class ConvWS2d(nn.Conv2d):
+
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 groups=1,
+                 bias=True,
+                 eps=1e-5):
+        super(ConvWS2d, self).__init__(
+            in_channels,
+            out_channels,
+            kernel_size,
+            stride=stride,
+            padding=padding,
+            dilation=dilation,
+            groups=groups,
+            bias=bias)
+        self.eps = eps
+
+    def forward(self, x):
+        return conv_ws_2d(x, self.weight, self.bias, self.stride, self.padding,
+                          self.dilation, self.groups, self.eps)
+
+
+@CONV_LAYERS.register_module(name='ConvAWS')
+class ConvAWS2d(nn.Conv2d):
+    """AWS (Adaptive Weight Standardization)
+
+    This is a variant of Weight Standardization
+    (https://arxiv.org/pdf/1903.10520.pdf)
+    It is used in DetectoRS to avoid NaN
+    (https://arxiv.org/pdf/2006.02334.pdf)
+
+    Args:
+        in_channels (int): Number of channels in the input image
+        out_channels (int): Number of channels produced by the convolution
+        kernel_size (int or tuple): Size of the conv kernel
+        stride (int or tuple, optional): Stride of the convolution. Default: 1
+        padding (int or tuple, optional): Zero-padding added to both sides of
+            the input. Default: 0
+        dilation (int or tuple, optional): Spacing between kernel elements.
+            Default: 1
+        groups (int, optional): Number of blocked connections from input
+            channels to output channels. Default: 1
+        bias (bool, optional): If set True, adds a learnable bias to the
+            output. Default: True
+    """
+
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 groups=1,
+                 bias=True):
+        super().__init__(
+            in_channels,
+            out_channels,
+            kernel_size,
+            stride=stride,
+            padding=padding,
+            dilation=dilation,
+            groups=groups,
+            bias=bias)
+        self.register_buffer('weight_gamma',
+                             torch.ones(self.out_channels, 1, 1, 1))
+        self.register_buffer('weight_beta',
+                             torch.zeros(self.out_channels, 1, 1, 1))
+
+    def _get_weight(self, weight):
+        weight_flat = weight.view(weight.size(0), -1)
+        mean = weight_flat.mean(dim=1).view(-1, 1, 1, 1)
+        std = torch.sqrt(weight_flat.var(dim=1) + 1e-5).view(-1, 1, 1, 1)
+        weight = (weight - mean) / std
+        weight = self.weight_gamma * weight + self.weight_beta
+        return weight
+
+    def forward(self, x):
+        weight = self._get_weight(self.weight)
+        return F.conv2d(x, weight, self.bias, self.stride, self.padding,
+                        self.dilation, self.groups)
+
+    def _load_from_state_dict(self, state_dict, prefix, local_metadata, strict,
+                              missing_keys, unexpected_keys, error_msgs):
+        """Override default load function.
+
+        AWS overrides the function _load_from_state_dict to recover
+        weight_gamma and weight_beta if they are missing. If weight_gamma and
+        weight_beta are found in the checkpoint, this function will return
+        after super()._load_from_state_dict. Otherwise, it will compute the
+        mean and std of the pretrained weights and store them in weight_beta
+        and weight_gamma.
+        """
+
+        self.weight_gamma.data.fill_(-1)
+        local_missing_keys = []
+        super()._load_from_state_dict(state_dict, prefix, local_metadata,
+                                      strict, local_missing_keys,
+                                      unexpected_keys, error_msgs)
+        if self.weight_gamma.data.mean() > 0:
+            for k in local_missing_keys:
+                missing_keys.append(k)
+            return
+        weight = self.weight.data
+        weight_flat = weight.view(weight.size(0), -1)
+        mean = weight_flat.mean(dim=1).view(-1, 1, 1, 1)
+        std = torch.sqrt(weight_flat.var(dim=1) + 1e-5).view(-1, 1, 1, 1)
+        self.weight_beta.data.copy_(mean)
+        self.weight_gamma.data.copy_(std)
+        missing_gamma_beta = [
+            k for k in local_missing_keys
+            if k.endswith('weight_gamma') or k.endswith('weight_beta')
+        ]
+        for k in missing_gamma_beta:
+            local_missing_keys.remove(k)
+        for k in local_missing_keys:
+            missing_keys.append(k)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/depthwise_separable_conv_module.py b/annotator/uniformer/mmcv/cnn/bricks/depthwise_separable_conv_module.py
new file mode 100644
index 0000000000000000000000000000000000000000..722d5d8d71f75486e2db3008907c4eadfca41d63
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/depthwise_separable_conv_module.py
@@ -0,0 +1,96 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch.nn as nn
+
+from .conv_module import ConvModule
+
+
+class DepthwiseSeparableConvModule(nn.Module):
+    """Depthwise separable convolution module.
+
+    See https://arxiv.org/pdf/1704.04861.pdf for details.
+
+    This module can replace a ConvModule with the conv block replaced by two
+    conv block: depthwise conv block and pointwise conv block. The depthwise
+    conv block contains depthwise-conv/norm/activation layers. The pointwise
+    conv block contains pointwise-conv/norm/activation layers. It should be
+    noted that there will be norm/activation layer in the depthwise conv block
+    if `norm_cfg` and `act_cfg` are specified.
+
+    Args:
+        in_channels (int): Number of channels in the input feature map.
+            Same as that in ``nn._ConvNd``.
+        out_channels (int): Number of channels produced by the convolution.
+            Same as that in ``nn._ConvNd``.
+        kernel_size (int | tuple[int]): Size of the convolving kernel.
+            Same as that in ``nn._ConvNd``.
+        stride (int | tuple[int]): Stride of the convolution.
+            Same as that in ``nn._ConvNd``. Default: 1.
+        padding (int | tuple[int]): Zero-padding added to both sides of
+            the input. Same as that in ``nn._ConvNd``. Default: 0.
+        dilation (int | tuple[int]): Spacing between kernel elements.
+            Same as that in ``nn._ConvNd``. Default: 1.
+        norm_cfg (dict): Default norm config for both depthwise ConvModule and
+            pointwise ConvModule. Default: None.
+        act_cfg (dict): Default activation config for both depthwise ConvModule
+            and pointwise ConvModule. Default: dict(type='ReLU').
+        dw_norm_cfg (dict): Norm config of depthwise ConvModule. If it is
+            'default', it will be the same as `norm_cfg`. Default: 'default'.
+        dw_act_cfg (dict): Activation config of depthwise ConvModule. If it is
+            'default', it will be the same as `act_cfg`. Default: 'default'.
+        pw_norm_cfg (dict): Norm config of pointwise ConvModule. If it is
+            'default', it will be the same as `norm_cfg`. Default: 'default'.
+        pw_act_cfg (dict): Activation config of pointwise ConvModule. If it is
+            'default', it will be the same as `act_cfg`. Default: 'default'.
+        kwargs (optional): Other shared arguments for depthwise and pointwise
+            ConvModule. See ConvModule for ref.
+    """
+
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 norm_cfg=None,
+                 act_cfg=dict(type='ReLU'),
+                 dw_norm_cfg='default',
+                 dw_act_cfg='default',
+                 pw_norm_cfg='default',
+                 pw_act_cfg='default',
+                 **kwargs):
+        super(DepthwiseSeparableConvModule, self).__init__()
+        assert 'groups' not in kwargs, 'groups should not be specified'
+
+        # if norm/activation config of depthwise/pointwise ConvModule is not
+        # specified, use default config.
+        dw_norm_cfg = dw_norm_cfg if dw_norm_cfg != 'default' else norm_cfg
+        dw_act_cfg = dw_act_cfg if dw_act_cfg != 'default' else act_cfg
+        pw_norm_cfg = pw_norm_cfg if pw_norm_cfg != 'default' else norm_cfg
+        pw_act_cfg = pw_act_cfg if pw_act_cfg != 'default' else act_cfg
+
+        # depthwise convolution
+        self.depthwise_conv = ConvModule(
+            in_channels,
+            in_channels,
+            kernel_size,
+            stride=stride,
+            padding=padding,
+            dilation=dilation,
+            groups=in_channels,
+            norm_cfg=dw_norm_cfg,
+            act_cfg=dw_act_cfg,
+            **kwargs)
+
+        self.pointwise_conv = ConvModule(
+            in_channels,
+            out_channels,
+            1,
+            norm_cfg=pw_norm_cfg,
+            act_cfg=pw_act_cfg,
+            **kwargs)
+
+    def forward(self, x):
+        x = self.depthwise_conv(x)
+        x = self.pointwise_conv(x)
+        return x
diff --git a/annotator/uniformer/mmcv/cnn/bricks/drop.py b/annotator/uniformer/mmcv/cnn/bricks/drop.py
new file mode 100644
index 0000000000000000000000000000000000000000..b7b4fccd457a0d51fb10c789df3c8537fe7b67c1
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/drop.py
@@ -0,0 +1,65 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+
+from annotator.uniformer.mmcv import build_from_cfg
+from .registry import DROPOUT_LAYERS
+
+
+def drop_path(x, drop_prob=0., training=False):
+    """Drop paths (Stochastic Depth) per sample (when applied in main path of
+    residual blocks).
+
+    We follow the implementation
+    https://github.com/rwightman/pytorch-image-models/blob/a2727c1bf78ba0d7b5727f5f95e37fb7f8866b1f/timm/models/layers/drop.py  # noqa: E501
+    """
+    if drop_prob == 0. or not training:
+        return x
+    keep_prob = 1 - drop_prob
+    # handle tensors with different dimensions, not just 4D tensors.
+    shape = (x.shape[0], ) + (1, ) * (x.ndim - 1)
+    random_tensor = keep_prob + torch.rand(
+        shape, dtype=x.dtype, device=x.device)
+    output = x.div(keep_prob) * random_tensor.floor()
+    return output
+
+
+@DROPOUT_LAYERS.register_module()
+class DropPath(nn.Module):
+    """Drop paths (Stochastic Depth) per sample  (when applied in main path of
+    residual blocks).
+
+    We follow the implementation
+    https://github.com/rwightman/pytorch-image-models/blob/a2727c1bf78ba0d7b5727f5f95e37fb7f8866b1f/timm/models/layers/drop.py  # noqa: E501
+
+    Args:
+        drop_prob (float): Probability of the path to be zeroed. Default: 0.1
+    """
+
+    def __init__(self, drop_prob=0.1):
+        super(DropPath, self).__init__()
+        self.drop_prob = drop_prob
+
+    def forward(self, x):
+        return drop_path(x, self.drop_prob, self.training)
+
+
+@DROPOUT_LAYERS.register_module()
+class Dropout(nn.Dropout):
+    """A wrapper for ``torch.nn.Dropout``, We rename the ``p`` of
+    ``torch.nn.Dropout`` to ``drop_prob`` so as to be consistent with
+    ``DropPath``
+
+    Args:
+        drop_prob (float): Probability of the elements to be
+            zeroed. Default: 0.5.
+        inplace (bool):  Do the operation inplace or not. Default: False.
+    """
+
+    def __init__(self, drop_prob=0.5, inplace=False):
+        super().__init__(p=drop_prob, inplace=inplace)
+
+
+def build_dropout(cfg, default_args=None):
+    """Builder for drop out layers."""
+    return build_from_cfg(cfg, DROPOUT_LAYERS, default_args)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/generalized_attention.py b/annotator/uniformer/mmcv/cnn/bricks/generalized_attention.py
new file mode 100644
index 0000000000000000000000000000000000000000..988d9adf2f289ef223bd1c680a5ae1d3387f0269
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/generalized_attention.py
@@ -0,0 +1,412 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import math
+
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from ..utils import kaiming_init
+from .registry import PLUGIN_LAYERS
+
+
+@PLUGIN_LAYERS.register_module()
+class GeneralizedAttention(nn.Module):
+    """GeneralizedAttention module.
+
+    See 'An Empirical Study of Spatial Attention Mechanisms in Deep Networks'
+    (https://arxiv.org/abs/1711.07971) for details.
+
+    Args:
+        in_channels (int): Channels of the input feature map.
+        spatial_range (int): The spatial range. -1 indicates no spatial range
+            constraint. Default: -1.
+        num_heads (int): The head number of empirical_attention module.
+            Default: 9.
+        position_embedding_dim (int): The position embedding dimension.
+            Default: -1.
+        position_magnitude (int): A multiplier acting on coord difference.
+            Default: 1.
+        kv_stride (int): The feature stride acting on key/value feature map.
+            Default: 2.
+        q_stride (int): The feature stride acting on query feature map.
+            Default: 1.
+        attention_type (str): A binary indicator string for indicating which
+            items in generalized empirical_attention module are used.
+            Default: '1111'.
+
+            - '1000' indicates 'query and key content' (appr - appr) item,
+            - '0100' indicates 'query content and relative position'
+              (appr - position) item,
+            - '0010' indicates 'key content only' (bias - appr) item,
+            - '0001' indicates 'relative position only' (bias - position) item.
+    """
+
+    _abbr_ = 'gen_attention_block'
+
+    def __init__(self,
+                 in_channels,
+                 spatial_range=-1,
+                 num_heads=9,
+                 position_embedding_dim=-1,
+                 position_magnitude=1,
+                 kv_stride=2,
+                 q_stride=1,
+                 attention_type='1111'):
+
+        super(GeneralizedAttention, self).__init__()
+
+        # hard range means local range for non-local operation
+        self.position_embedding_dim = (
+            position_embedding_dim
+            if position_embedding_dim > 0 else in_channels)
+
+        self.position_magnitude = position_magnitude
+        self.num_heads = num_heads
+        self.in_channels = in_channels
+        self.spatial_range = spatial_range
+        self.kv_stride = kv_stride
+        self.q_stride = q_stride
+        self.attention_type = [bool(int(_)) for _ in attention_type]
+        self.qk_embed_dim = in_channels // num_heads
+        out_c = self.qk_embed_dim * num_heads
+
+        if self.attention_type[0] or self.attention_type[1]:
+            self.query_conv = nn.Conv2d(
+                in_channels=in_channels,
+                out_channels=out_c,
+                kernel_size=1,
+                bias=False)
+            self.query_conv.kaiming_init = True
+
+        if self.attention_type[0] or self.attention_type[2]:
+            self.key_conv = nn.Conv2d(
+                in_channels=in_channels,
+                out_channels=out_c,
+                kernel_size=1,
+                bias=False)
+            self.key_conv.kaiming_init = True
+
+        self.v_dim = in_channels // num_heads
+        self.value_conv = nn.Conv2d(
+            in_channels=in_channels,
+            out_channels=self.v_dim * num_heads,
+            kernel_size=1,
+            bias=False)
+        self.value_conv.kaiming_init = True
+
+        if self.attention_type[1] or self.attention_type[3]:
+            self.appr_geom_fc_x = nn.Linear(
+                self.position_embedding_dim // 2, out_c, bias=False)
+            self.appr_geom_fc_x.kaiming_init = True
+
+            self.appr_geom_fc_y = nn.Linear(
+                self.position_embedding_dim // 2, out_c, bias=False)
+            self.appr_geom_fc_y.kaiming_init = True
+
+        if self.attention_type[2]:
+            stdv = 1.0 / math.sqrt(self.qk_embed_dim * 2)
+            appr_bias_value = -2 * stdv * torch.rand(out_c) + stdv
+            self.appr_bias = nn.Parameter(appr_bias_value)
+
+        if self.attention_type[3]:
+            stdv = 1.0 / math.sqrt(self.qk_embed_dim * 2)
+            geom_bias_value = -2 * stdv * torch.rand(out_c) + stdv
+            self.geom_bias = nn.Parameter(geom_bias_value)
+
+        self.proj_conv = nn.Conv2d(
+            in_channels=self.v_dim * num_heads,
+            out_channels=in_channels,
+            kernel_size=1,
+            bias=True)
+        self.proj_conv.kaiming_init = True
+        self.gamma = nn.Parameter(torch.zeros(1))
+
+        if self.spatial_range >= 0:
+            # only works when non local is after 3*3 conv
+            if in_channels == 256:
+                max_len = 84
+            elif in_channels == 512:
+                max_len = 42
+
+            max_len_kv = int((max_len - 1.0) / self.kv_stride + 1)
+            local_constraint_map = np.ones(
+                (max_len, max_len, max_len_kv, max_len_kv), dtype=np.int)
+            for iy in range(max_len):
+                for ix in range(max_len):
+                    local_constraint_map[
+                        iy, ix,
+                        max((iy - self.spatial_range) //
+                            self.kv_stride, 0):min((iy + self.spatial_range +
+                                                    1) // self.kv_stride +
+                                                   1, max_len),
+                        max((ix - self.spatial_range) //
+                            self.kv_stride, 0):min((ix + self.spatial_range +
+                                                    1) // self.kv_stride +
+                                                   1, max_len)] = 0
+
+            self.local_constraint_map = nn.Parameter(
+                torch.from_numpy(local_constraint_map).byte(),
+                requires_grad=False)
+
+        if self.q_stride > 1:
+            self.q_downsample = nn.AvgPool2d(
+                kernel_size=1, stride=self.q_stride)
+        else:
+            self.q_downsample = None
+
+        if self.kv_stride > 1:
+            self.kv_downsample = nn.AvgPool2d(
+                kernel_size=1, stride=self.kv_stride)
+        else:
+            self.kv_downsample = None
+
+        self.init_weights()
+
+    def get_position_embedding(self,
+                               h,
+                               w,
+                               h_kv,
+                               w_kv,
+                               q_stride,
+                               kv_stride,
+                               device,
+                               dtype,
+                               feat_dim,
+                               wave_length=1000):
+        # the default type of Tensor is float32, leading to type mismatch
+        # in fp16 mode. Cast it to support fp16 mode.
+        h_idxs = torch.linspace(0, h - 1, h).to(device=device, dtype=dtype)
+        h_idxs = h_idxs.view((h, 1)) * q_stride
+
+        w_idxs = torch.linspace(0, w - 1, w).to(device=device, dtype=dtype)
+        w_idxs = w_idxs.view((w, 1)) * q_stride
+
+        h_kv_idxs = torch.linspace(0, h_kv - 1, h_kv).to(
+            device=device, dtype=dtype)
+        h_kv_idxs = h_kv_idxs.view((h_kv, 1)) * kv_stride
+
+        w_kv_idxs = torch.linspace(0, w_kv - 1, w_kv).to(
+            device=device, dtype=dtype)
+        w_kv_idxs = w_kv_idxs.view((w_kv, 1)) * kv_stride
+
+        # (h, h_kv, 1)
+        h_diff = h_idxs.unsqueeze(1) - h_kv_idxs.unsqueeze(0)
+        h_diff *= self.position_magnitude
+
+        # (w, w_kv, 1)
+        w_diff = w_idxs.unsqueeze(1) - w_kv_idxs.unsqueeze(0)
+        w_diff *= self.position_magnitude
+
+        feat_range = torch.arange(0, feat_dim / 4).to(
+            device=device, dtype=dtype)
+
+        dim_mat = torch.Tensor([wave_length]).to(device=device, dtype=dtype)
+        dim_mat = dim_mat**((4. / feat_dim) * feat_range)
+        dim_mat = dim_mat.view((1, 1, -1))
+
+        embedding_x = torch.cat(
+            ((w_diff / dim_mat).sin(), (w_diff / dim_mat).cos()), dim=2)
+
+        embedding_y = torch.cat(
+            ((h_diff / dim_mat).sin(), (h_diff / dim_mat).cos()), dim=2)
+
+        return embedding_x, embedding_y
+
+    def forward(self, x_input):
+        num_heads = self.num_heads
+
+        # use empirical_attention
+        if self.q_downsample is not None:
+            x_q = self.q_downsample(x_input)
+        else:
+            x_q = x_input
+        n, _, h, w = x_q.shape
+
+        if self.kv_downsample is not None:
+            x_kv = self.kv_downsample(x_input)
+        else:
+            x_kv = x_input
+        _, _, h_kv, w_kv = x_kv.shape
+
+        if self.attention_type[0] or self.attention_type[1]:
+            proj_query = self.query_conv(x_q).view(
+                (n, num_heads, self.qk_embed_dim, h * w))
+            proj_query = proj_query.permute(0, 1, 3, 2)
+
+        if self.attention_type[0] or self.attention_type[2]:
+            proj_key = self.key_conv(x_kv).view(
+                (n, num_heads, self.qk_embed_dim, h_kv * w_kv))
+
+        if self.attention_type[1] or self.attention_type[3]:
+            position_embed_x, position_embed_y = self.get_position_embedding(
+                h, w, h_kv, w_kv, self.q_stride, self.kv_stride,
+                x_input.device, x_input.dtype, self.position_embedding_dim)
+            # (n, num_heads, w, w_kv, dim)
+            position_feat_x = self.appr_geom_fc_x(position_embed_x).\
+                view(1, w, w_kv, num_heads, self.qk_embed_dim).\
+                permute(0, 3, 1, 2, 4).\
+                repeat(n, 1, 1, 1, 1)
+
+            # (n, num_heads, h, h_kv, dim)
+            position_feat_y = self.appr_geom_fc_y(position_embed_y).\
+                view(1, h, h_kv, num_heads, self.qk_embed_dim).\
+                permute(0, 3, 1, 2, 4).\
+                repeat(n, 1, 1, 1, 1)
+
+            position_feat_x /= math.sqrt(2)
+            position_feat_y /= math.sqrt(2)
+
+        # accelerate for saliency only
+        if (np.sum(self.attention_type) == 1) and self.attention_type[2]:
+            appr_bias = self.appr_bias.\
+                view(1, num_heads, 1, self.qk_embed_dim).\
+                repeat(n, 1, 1, 1)
+
+            energy = torch.matmul(appr_bias, proj_key).\
+                view(n, num_heads, 1, h_kv * w_kv)
+
+            h = 1
+            w = 1
+        else:
+            # (n, num_heads, h*w, h_kv*w_kv), query before key, 540mb for
+            if not self.attention_type[0]:
+                energy = torch.zeros(
+                    n,
+                    num_heads,
+                    h,
+                    w,
+                    h_kv,
+                    w_kv,
+                    dtype=x_input.dtype,
+                    device=x_input.device)
+
+            # attention_type[0]: appr - appr
+            # attention_type[1]: appr - position
+            # attention_type[2]: bias - appr
+            # attention_type[3]: bias - position
+            if self.attention_type[0] or self.attention_type[2]:
+                if self.attention_type[0] and self.attention_type[2]:
+                    appr_bias = self.appr_bias.\
+                        view(1, num_heads, 1, self.qk_embed_dim)
+                    energy = torch.matmul(proj_query + appr_bias, proj_key).\
+                        view(n, num_heads, h, w, h_kv, w_kv)
+
+                elif self.attention_type[0]:
+                    energy = torch.matmul(proj_query, proj_key).\
+                        view(n, num_heads, h, w, h_kv, w_kv)
+
+                elif self.attention_type[2]:
+                    appr_bias = self.appr_bias.\
+                        view(1, num_heads, 1, self.qk_embed_dim).\
+                        repeat(n, 1, 1, 1)
+
+                    energy += torch.matmul(appr_bias, proj_key).\
+                        view(n, num_heads, 1, 1, h_kv, w_kv)
+
+            if self.attention_type[1] or self.attention_type[3]:
+                if self.attention_type[1] and self.attention_type[3]:
+                    geom_bias = self.geom_bias.\
+                        view(1, num_heads, 1, self.qk_embed_dim)
+
+                    proj_query_reshape = (proj_query + geom_bias).\
+                        view(n, num_heads, h, w, self.qk_embed_dim)
+
+                    energy_x = torch.matmul(
+                        proj_query_reshape.permute(0, 1, 3, 2, 4),
+                        position_feat_x.permute(0, 1, 2, 4, 3))
+                    energy_x = energy_x.\
+                        permute(0, 1, 3, 2, 4).unsqueeze(4)
+
+                    energy_y = torch.matmul(
+                        proj_query_reshape,
+                        position_feat_y.permute(0, 1, 2, 4, 3))
+                    energy_y = energy_y.unsqueeze(5)
+
+                    energy += energy_x + energy_y
+
+                elif self.attention_type[1]:
+                    proj_query_reshape = proj_query.\
+                        view(n, num_heads, h, w, self.qk_embed_dim)
+                    proj_query_reshape = proj_query_reshape.\
+                        permute(0, 1, 3, 2, 4)
+                    position_feat_x_reshape = position_feat_x.\
+                        permute(0, 1, 2, 4, 3)
+                    position_feat_y_reshape = position_feat_y.\
+                        permute(0, 1, 2, 4, 3)
+
+                    energy_x = torch.matmul(proj_query_reshape,
+                                            position_feat_x_reshape)
+                    energy_x = energy_x.permute(0, 1, 3, 2, 4).unsqueeze(4)
+
+                    energy_y = torch.matmul(proj_query_reshape,
+                                            position_feat_y_reshape)
+                    energy_y = energy_y.unsqueeze(5)
+
+                    energy += energy_x + energy_y
+
+                elif self.attention_type[3]:
+                    geom_bias = self.geom_bias.\
+                        view(1, num_heads, self.qk_embed_dim, 1).\
+                        repeat(n, 1, 1, 1)
+
+                    position_feat_x_reshape = position_feat_x.\
+                        view(n, num_heads, w*w_kv, self.qk_embed_dim)
+
+                    position_feat_y_reshape = position_feat_y.\
+                        view(n, num_heads, h * h_kv, self.qk_embed_dim)
+
+                    energy_x = torch.matmul(position_feat_x_reshape, geom_bias)
+                    energy_x = energy_x.view(n, num_heads, 1, w, 1, w_kv)
+
+                    energy_y = torch.matmul(position_feat_y_reshape, geom_bias)
+                    energy_y = energy_y.view(n, num_heads, h, 1, h_kv, 1)
+
+                    energy += energy_x + energy_y
+
+            energy = energy.view(n, num_heads, h * w, h_kv * w_kv)
+
+        if self.spatial_range >= 0:
+            cur_local_constraint_map = \
+                self.local_constraint_map[:h, :w, :h_kv, :w_kv].\
+                contiguous().\
+                view(1, 1, h*w, h_kv*w_kv)
+
+            energy = energy.masked_fill_(cur_local_constraint_map,
+                                         float('-inf'))
+
+        attention = F.softmax(energy, 3)
+
+        proj_value = self.value_conv(x_kv)
+        proj_value_reshape = proj_value.\
+            view((n, num_heads, self.v_dim, h_kv * w_kv)).\
+            permute(0, 1, 3, 2)
+
+        out = torch.matmul(attention, proj_value_reshape).\
+            permute(0, 1, 3, 2).\
+            contiguous().\
+            view(n, self.v_dim * self.num_heads, h, w)
+
+        out = self.proj_conv(out)
+
+        # output is downsampled, upsample back to input size
+        if self.q_downsample is not None:
+            out = F.interpolate(
+                out,
+                size=x_input.shape[2:],
+                mode='bilinear',
+                align_corners=False)
+
+        out = self.gamma * out + x_input
+        return out
+
+    def init_weights(self):
+        for m in self.modules():
+            if hasattr(m, 'kaiming_init') and m.kaiming_init:
+                kaiming_init(
+                    m,
+                    mode='fan_in',
+                    nonlinearity='leaky_relu',
+                    bias=0,
+                    distribution='uniform',
+                    a=1)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/hsigmoid.py b/annotator/uniformer/mmcv/cnn/bricks/hsigmoid.py
new file mode 100644
index 0000000000000000000000000000000000000000..30b1a3d6580cf0360710426fbea1f05acdf07b4b
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/hsigmoid.py
@@ -0,0 +1,34 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch.nn as nn
+
+from .registry import ACTIVATION_LAYERS
+
+
+@ACTIVATION_LAYERS.register_module()
+class HSigmoid(nn.Module):
+    """Hard Sigmoid Module. Apply the hard sigmoid function:
+    Hsigmoid(x) = min(max((x + bias) / divisor, min_value), max_value)
+    Default: Hsigmoid(x) = min(max((x + 1) / 2, 0), 1)
+
+    Args:
+        bias (float): Bias of the input feature map. Default: 1.0.
+        divisor (float): Divisor of the input feature map. Default: 2.0.
+        min_value (float): Lower bound value. Default: 0.0.
+        max_value (float): Upper bound value. Default: 1.0.
+
+    Returns:
+        Tensor: The output tensor.
+    """
+
+    def __init__(self, bias=1.0, divisor=2.0, min_value=0.0, max_value=1.0):
+        super(HSigmoid, self).__init__()
+        self.bias = bias
+        self.divisor = divisor
+        assert self.divisor != 0
+        self.min_value = min_value
+        self.max_value = max_value
+
+    def forward(self, x):
+        x = (x + self.bias) / self.divisor
+
+        return x.clamp_(self.min_value, self.max_value)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/hswish.py b/annotator/uniformer/mmcv/cnn/bricks/hswish.py
new file mode 100644
index 0000000000000000000000000000000000000000..7e0c090ff037c99ee6c5c84c4592e87beae02208
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/hswish.py
@@ -0,0 +1,29 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch.nn as nn
+
+from .registry import ACTIVATION_LAYERS
+
+
+@ACTIVATION_LAYERS.register_module()
+class HSwish(nn.Module):
+    """Hard Swish Module.
+
+    This module applies the hard swish function:
+
+    .. math::
+        Hswish(x) = x * ReLU6(x + 3) / 6
+
+    Args:
+        inplace (bool): can optionally do the operation in-place.
+            Default: False.
+
+    Returns:
+        Tensor: The output tensor.
+    """
+
+    def __init__(self, inplace=False):
+        super(HSwish, self).__init__()
+        self.act = nn.ReLU6(inplace)
+
+    def forward(self, x):
+        return x * self.act(x + 3) / 6
diff --git a/annotator/uniformer/mmcv/cnn/bricks/non_local.py b/annotator/uniformer/mmcv/cnn/bricks/non_local.py
new file mode 100644
index 0000000000000000000000000000000000000000..92d00155ef275c1201ea66bba30470a1785cc5d7
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/non_local.py
@@ -0,0 +1,306 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from abc import ABCMeta
+
+import torch
+import torch.nn as nn
+
+from ..utils import constant_init, normal_init
+from .conv_module import ConvModule
+from .registry import PLUGIN_LAYERS
+
+
+class _NonLocalNd(nn.Module, metaclass=ABCMeta):
+    """Basic Non-local module.
+
+    This module is proposed in
+    "Non-local Neural Networks"
+    Paper reference: https://arxiv.org/abs/1711.07971
+    Code reference: https://github.com/AlexHex7/Non-local_pytorch
+
+    Args:
+        in_channels (int): Channels of the input feature map.
+        reduction (int): Channel reduction ratio. Default: 2.
+        use_scale (bool): Whether to scale pairwise_weight by
+            `1/sqrt(inter_channels)` when the mode is `embedded_gaussian`.
+            Default: True.
+        conv_cfg (None | dict): The config dict for convolution layers.
+            If not specified, it will use `nn.Conv2d` for convolution layers.
+            Default: None.
+        norm_cfg (None | dict): The config dict for normalization layers.
+            Default: None. (This parameter is only applicable to conv_out.)
+        mode (str): Options are `gaussian`, `concatenation`,
+            `embedded_gaussian` and `dot_product`. Default: embedded_gaussian.
+    """
+
+    def __init__(self,
+                 in_channels,
+                 reduction=2,
+                 use_scale=True,
+                 conv_cfg=None,
+                 norm_cfg=None,
+                 mode='embedded_gaussian',
+                 **kwargs):
+        super(_NonLocalNd, self).__init__()
+        self.in_channels = in_channels
+        self.reduction = reduction
+        self.use_scale = use_scale
+        self.inter_channels = max(in_channels // reduction, 1)
+        self.mode = mode
+
+        if mode not in [
+                'gaussian', 'embedded_gaussian', 'dot_product', 'concatenation'
+        ]:
+            raise ValueError("Mode should be in 'gaussian', 'concatenation', "
+                             f"'embedded_gaussian' or 'dot_product', but got "
+                             f'{mode} instead.')
+
+        # g, theta, phi are defaulted as `nn.ConvNd`.
+        # Here we use ConvModule for potential usage.
+        self.g = ConvModule(
+            self.in_channels,
+            self.inter_channels,
+            kernel_size=1,
+            conv_cfg=conv_cfg,
+            act_cfg=None)
+        self.conv_out = ConvModule(
+            self.inter_channels,
+            self.in_channels,
+            kernel_size=1,
+            conv_cfg=conv_cfg,
+            norm_cfg=norm_cfg,
+            act_cfg=None)
+
+        if self.mode != 'gaussian':
+            self.theta = ConvModule(
+                self.in_channels,
+                self.inter_channels,
+                kernel_size=1,
+                conv_cfg=conv_cfg,
+                act_cfg=None)
+            self.phi = ConvModule(
+                self.in_channels,
+                self.inter_channels,
+                kernel_size=1,
+                conv_cfg=conv_cfg,
+                act_cfg=None)
+
+        if self.mode == 'concatenation':
+            self.concat_project = ConvModule(
+                self.inter_channels * 2,
+                1,
+                kernel_size=1,
+                stride=1,
+                padding=0,
+                bias=False,
+                act_cfg=dict(type='ReLU'))
+
+        self.init_weights(**kwargs)
+
+    def init_weights(self, std=0.01, zeros_init=True):
+        if self.mode != 'gaussian':
+            for m in [self.g, self.theta, self.phi]:
+                normal_init(m.conv, std=std)
+        else:
+            normal_init(self.g.conv, std=std)
+        if zeros_init:
+            if self.conv_out.norm_cfg is None:
+                constant_init(self.conv_out.conv, 0)
+            else:
+                constant_init(self.conv_out.norm, 0)
+        else:
+            if self.conv_out.norm_cfg is None:
+                normal_init(self.conv_out.conv, std=std)
+            else:
+                normal_init(self.conv_out.norm, std=std)
+
+    def gaussian(self, theta_x, phi_x):
+        # NonLocal1d pairwise_weight: [N, H, H]
+        # NonLocal2d pairwise_weight: [N, HxW, HxW]
+        # NonLocal3d pairwise_weight: [N, TxHxW, TxHxW]
+        pairwise_weight = torch.matmul(theta_x, phi_x)
+        pairwise_weight = pairwise_weight.softmax(dim=-1)
+        return pairwise_weight
+
+    def embedded_gaussian(self, theta_x, phi_x):
+        # NonLocal1d pairwise_weight: [N, H, H]
+        # NonLocal2d pairwise_weight: [N, HxW, HxW]
+        # NonLocal3d pairwise_weight: [N, TxHxW, TxHxW]
+        pairwise_weight = torch.matmul(theta_x, phi_x)
+        if self.use_scale:
+            # theta_x.shape[-1] is `self.inter_channels`
+            pairwise_weight /= theta_x.shape[-1]**0.5
+        pairwise_weight = pairwise_weight.softmax(dim=-1)
+        return pairwise_weight
+
+    def dot_product(self, theta_x, phi_x):
+        # NonLocal1d pairwise_weight: [N, H, H]
+        # NonLocal2d pairwise_weight: [N, HxW, HxW]
+        # NonLocal3d pairwise_weight: [N, TxHxW, TxHxW]
+        pairwise_weight = torch.matmul(theta_x, phi_x)
+        pairwise_weight /= pairwise_weight.shape[-1]
+        return pairwise_weight
+
+    def concatenation(self, theta_x, phi_x):
+        # NonLocal1d pairwise_weight: [N, H, H]
+        # NonLocal2d pairwise_weight: [N, HxW, HxW]
+        # NonLocal3d pairwise_weight: [N, TxHxW, TxHxW]
+        h = theta_x.size(2)
+        w = phi_x.size(3)
+        theta_x = theta_x.repeat(1, 1, 1, w)
+        phi_x = phi_x.repeat(1, 1, h, 1)
+
+        concat_feature = torch.cat([theta_x, phi_x], dim=1)
+        pairwise_weight = self.concat_project(concat_feature)
+        n, _, h, w = pairwise_weight.size()
+        pairwise_weight = pairwise_weight.view(n, h, w)
+        pairwise_weight /= pairwise_weight.shape[-1]
+
+        return pairwise_weight
+
+    def forward(self, x):
+        # Assume `reduction = 1`, then `inter_channels = C`
+        # or `inter_channels = C` when `mode="gaussian"`
+
+        # NonLocal1d x: [N, C, H]
+        # NonLocal2d x: [N, C, H, W]
+        # NonLocal3d x: [N, C, T, H, W]
+        n = x.size(0)
+
+        # NonLocal1d g_x: [N, H, C]
+        # NonLocal2d g_x: [N, HxW, C]
+        # NonLocal3d g_x: [N, TxHxW, C]
+        g_x = self.g(x).view(n, self.inter_channels, -1)
+        g_x = g_x.permute(0, 2, 1)
+
+        # NonLocal1d theta_x: [N, H, C], phi_x: [N, C, H]
+        # NonLocal2d theta_x: [N, HxW, C], phi_x: [N, C, HxW]
+        # NonLocal3d theta_x: [N, TxHxW, C], phi_x: [N, C, TxHxW]
+        if self.mode == 'gaussian':
+            theta_x = x.view(n, self.in_channels, -1)
+            theta_x = theta_x.permute(0, 2, 1)
+            if self.sub_sample:
+                phi_x = self.phi(x).view(n, self.in_channels, -1)
+            else:
+                phi_x = x.view(n, self.in_channels, -1)
+        elif self.mode == 'concatenation':
+            theta_x = self.theta(x).view(n, self.inter_channels, -1, 1)
+            phi_x = self.phi(x).view(n, self.inter_channels, 1, -1)
+        else:
+            theta_x = self.theta(x).view(n, self.inter_channels, -1)
+            theta_x = theta_x.permute(0, 2, 1)
+            phi_x = self.phi(x).view(n, self.inter_channels, -1)
+
+        pairwise_func = getattr(self, self.mode)
+        # NonLocal1d pairwise_weight: [N, H, H]
+        # NonLocal2d pairwise_weight: [N, HxW, HxW]
+        # NonLocal3d pairwise_weight: [N, TxHxW, TxHxW]
+        pairwise_weight = pairwise_func(theta_x, phi_x)
+
+        # NonLocal1d y: [N, H, C]
+        # NonLocal2d y: [N, HxW, C]
+        # NonLocal3d y: [N, TxHxW, C]
+        y = torch.matmul(pairwise_weight, g_x)
+        # NonLocal1d y: [N, C, H]
+        # NonLocal2d y: [N, C, H, W]
+        # NonLocal3d y: [N, C, T, H, W]
+        y = y.permute(0, 2, 1).contiguous().reshape(n, self.inter_channels,
+                                                    *x.size()[2:])
+
+        output = x + self.conv_out(y)
+
+        return output
+
+
+class NonLocal1d(_NonLocalNd):
+    """1D Non-local module.
+
+    Args:
+        in_channels (int): Same as `NonLocalND`.
+        sub_sample (bool): Whether to apply max pooling after pairwise
+            function (Note that the `sub_sample` is applied on spatial only).
+            Default: False.
+        conv_cfg (None | dict): Same as `NonLocalND`.
+            Default: dict(type='Conv1d').
+    """
+
+    def __init__(self,
+                 in_channels,
+                 sub_sample=False,
+                 conv_cfg=dict(type='Conv1d'),
+                 **kwargs):
+        super(NonLocal1d, self).__init__(
+            in_channels, conv_cfg=conv_cfg, **kwargs)
+
+        self.sub_sample = sub_sample
+
+        if sub_sample:
+            max_pool_layer = nn.MaxPool1d(kernel_size=2)
+            self.g = nn.Sequential(self.g, max_pool_layer)
+            if self.mode != 'gaussian':
+                self.phi = nn.Sequential(self.phi, max_pool_layer)
+            else:
+                self.phi = max_pool_layer
+
+
+@PLUGIN_LAYERS.register_module()
+class NonLocal2d(_NonLocalNd):
+    """2D Non-local module.
+
+    Args:
+        in_channels (int): Same as `NonLocalND`.
+        sub_sample (bool): Whether to apply max pooling after pairwise
+            function (Note that the `sub_sample` is applied on spatial only).
+            Default: False.
+        conv_cfg (None | dict): Same as `NonLocalND`.
+            Default: dict(type='Conv2d').
+    """
+
+    _abbr_ = 'nonlocal_block'
+
+    def __init__(self,
+                 in_channels,
+                 sub_sample=False,
+                 conv_cfg=dict(type='Conv2d'),
+                 **kwargs):
+        super(NonLocal2d, self).__init__(
+            in_channels, conv_cfg=conv_cfg, **kwargs)
+
+        self.sub_sample = sub_sample
+
+        if sub_sample:
+            max_pool_layer = nn.MaxPool2d(kernel_size=(2, 2))
+            self.g = nn.Sequential(self.g, max_pool_layer)
+            if self.mode != 'gaussian':
+                self.phi = nn.Sequential(self.phi, max_pool_layer)
+            else:
+                self.phi = max_pool_layer
+
+
+class NonLocal3d(_NonLocalNd):
+    """3D Non-local module.
+
+    Args:
+        in_channels (int): Same as `NonLocalND`.
+        sub_sample (bool): Whether to apply max pooling after pairwise
+            function (Note that the `sub_sample` is applied on spatial only).
+            Default: False.
+        conv_cfg (None | dict): Same as `NonLocalND`.
+            Default: dict(type='Conv3d').
+    """
+
+    def __init__(self,
+                 in_channels,
+                 sub_sample=False,
+                 conv_cfg=dict(type='Conv3d'),
+                 **kwargs):
+        super(NonLocal3d, self).__init__(
+            in_channels, conv_cfg=conv_cfg, **kwargs)
+        self.sub_sample = sub_sample
+
+        if sub_sample:
+            max_pool_layer = nn.MaxPool3d(kernel_size=(1, 2, 2))
+            self.g = nn.Sequential(self.g, max_pool_layer)
+            if self.mode != 'gaussian':
+                self.phi = nn.Sequential(self.phi, max_pool_layer)
+            else:
+                self.phi = max_pool_layer
diff --git a/annotator/uniformer/mmcv/cnn/bricks/norm.py b/annotator/uniformer/mmcv/cnn/bricks/norm.py
new file mode 100644
index 0000000000000000000000000000000000000000..408f4b42731b19a3beeef68b6a5e610d0bbc18b3
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/norm.py
@@ -0,0 +1,144 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import inspect
+
+import torch.nn as nn
+
+from annotator.uniformer.mmcv.utils import is_tuple_of
+from annotator.uniformer.mmcv.utils.parrots_wrapper import SyncBatchNorm, _BatchNorm, _InstanceNorm
+from .registry import NORM_LAYERS
+
+NORM_LAYERS.register_module('BN', module=nn.BatchNorm2d)
+NORM_LAYERS.register_module('BN1d', module=nn.BatchNorm1d)
+NORM_LAYERS.register_module('BN2d', module=nn.BatchNorm2d)
+NORM_LAYERS.register_module('BN3d', module=nn.BatchNorm3d)
+NORM_LAYERS.register_module('SyncBN', module=SyncBatchNorm)
+NORM_LAYERS.register_module('GN', module=nn.GroupNorm)
+NORM_LAYERS.register_module('LN', module=nn.LayerNorm)
+NORM_LAYERS.register_module('IN', module=nn.InstanceNorm2d)
+NORM_LAYERS.register_module('IN1d', module=nn.InstanceNorm1d)
+NORM_LAYERS.register_module('IN2d', module=nn.InstanceNorm2d)
+NORM_LAYERS.register_module('IN3d', module=nn.InstanceNorm3d)
+
+
+def infer_abbr(class_type):
+    """Infer abbreviation from the class name.
+
+    When we build a norm layer with `build_norm_layer()`, we want to preserve
+    the norm type in variable names, e.g, self.bn1, self.gn. This method will
+    infer the abbreviation to map class types to abbreviations.
+
+    Rule 1: If the class has the property "_abbr_", return the property.
+    Rule 2: If the parent class is _BatchNorm, GroupNorm, LayerNorm or
+    InstanceNorm, the abbreviation of this layer will be "bn", "gn", "ln" and
+    "in" respectively.
+    Rule 3: If the class name contains "batch", "group", "layer" or "instance",
+    the abbreviation of this layer will be "bn", "gn", "ln" and "in"
+    respectively.
+    Rule 4: Otherwise, the abbreviation falls back to "norm".
+
+    Args:
+        class_type (type): The norm layer type.
+
+    Returns:
+        str: The inferred abbreviation.
+    """
+    if not inspect.isclass(class_type):
+        raise TypeError(
+            f'class_type must be a type, but got {type(class_type)}')
+    if hasattr(class_type, '_abbr_'):
+        return class_type._abbr_
+    if issubclass(class_type, _InstanceNorm):  # IN is a subclass of BN
+        return 'in'
+    elif issubclass(class_type, _BatchNorm):
+        return 'bn'
+    elif issubclass(class_type, nn.GroupNorm):
+        return 'gn'
+    elif issubclass(class_type, nn.LayerNorm):
+        return 'ln'
+    else:
+        class_name = class_type.__name__.lower()
+        if 'batch' in class_name:
+            return 'bn'
+        elif 'group' in class_name:
+            return 'gn'
+        elif 'layer' in class_name:
+            return 'ln'
+        elif 'instance' in class_name:
+            return 'in'
+        else:
+            return 'norm_layer'
+
+
+def build_norm_layer(cfg, num_features, postfix=''):
+    """Build normalization layer.
+
+    Args:
+        cfg (dict): The norm layer config, which should contain:
+
+            - type (str): Layer type.
+            - layer args: Args needed to instantiate a norm layer.
+            - requires_grad (bool, optional): Whether stop gradient updates.
+        num_features (int): Number of input channels.
+        postfix (int | str): The postfix to be appended into norm abbreviation
+            to create named layer.
+
+    Returns:
+        (str, nn.Module): The first element is the layer name consisting of
+            abbreviation and postfix, e.g., bn1, gn. The second element is the
+            created norm layer.
+    """
+    if not isinstance(cfg, dict):
+        raise TypeError('cfg must be a dict')
+    if 'type' not in cfg:
+        raise KeyError('the cfg dict must contain the key "type"')
+    cfg_ = cfg.copy()
+
+    layer_type = cfg_.pop('type')
+    if layer_type not in NORM_LAYERS:
+        raise KeyError(f'Unrecognized norm type {layer_type}')
+
+    norm_layer = NORM_LAYERS.get(layer_type)
+    abbr = infer_abbr(norm_layer)
+
+    assert isinstance(postfix, (int, str))
+    name = abbr + str(postfix)
+
+    requires_grad = cfg_.pop('requires_grad', True)
+    cfg_.setdefault('eps', 1e-5)
+    if layer_type != 'GN':
+        layer = norm_layer(num_features, **cfg_)
+        if layer_type == 'SyncBN' and hasattr(layer, '_specify_ddp_gpu_num'):
+            layer._specify_ddp_gpu_num(1)
+    else:
+        assert 'num_groups' in cfg_
+        layer = norm_layer(num_channels=num_features, **cfg_)
+
+    for param in layer.parameters():
+        param.requires_grad = requires_grad
+
+    return name, layer
+
+
+def is_norm(layer, exclude=None):
+    """Check if a layer is a normalization layer.
+
+    Args:
+        layer (nn.Module): The layer to be checked.
+        exclude (type | tuple[type]): Types to be excluded.
+
+    Returns:
+        bool: Whether the layer is a norm layer.
+    """
+    if exclude is not None:
+        if not isinstance(exclude, tuple):
+            exclude = (exclude, )
+        if not is_tuple_of(exclude, type):
+            raise TypeError(
+                f'"exclude" must be either None or type or a tuple of types, '
+                f'but got {type(exclude)}: {exclude}')
+
+    if exclude and isinstance(layer, exclude):
+        return False
+
+    all_norm_bases = (_BatchNorm, _InstanceNorm, nn.GroupNorm, nn.LayerNorm)
+    return isinstance(layer, all_norm_bases)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/padding.py b/annotator/uniformer/mmcv/cnn/bricks/padding.py
new file mode 100644
index 0000000000000000000000000000000000000000..e4ac6b28a1789bd551c613a7d3e7b622433ac7ec
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/padding.py
@@ -0,0 +1,36 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch.nn as nn
+
+from .registry import PADDING_LAYERS
+
+PADDING_LAYERS.register_module('zero', module=nn.ZeroPad2d)
+PADDING_LAYERS.register_module('reflect', module=nn.ReflectionPad2d)
+PADDING_LAYERS.register_module('replicate', module=nn.ReplicationPad2d)
+
+
+def build_padding_layer(cfg, *args, **kwargs):
+    """Build padding layer.
+
+    Args:
+        cfg (None or dict): The padding layer config, which should contain:
+            - type (str): Layer type.
+            - layer args: Args needed to instantiate a padding layer.
+
+    Returns:
+        nn.Module: Created padding layer.
+    """
+    if not isinstance(cfg, dict):
+        raise TypeError('cfg must be a dict')
+    if 'type' not in cfg:
+        raise KeyError('the cfg dict must contain the key "type"')
+
+    cfg_ = cfg.copy()
+    padding_type = cfg_.pop('type')
+    if padding_type not in PADDING_LAYERS:
+        raise KeyError(f'Unrecognized padding type {padding_type}.')
+    else:
+        padding_layer = PADDING_LAYERS.get(padding_type)
+
+    layer = padding_layer(*args, **kwargs, **cfg_)
+
+    return layer
diff --git a/annotator/uniformer/mmcv/cnn/bricks/plugin.py b/annotator/uniformer/mmcv/cnn/bricks/plugin.py
new file mode 100644
index 0000000000000000000000000000000000000000..07c010d4053174dd41107aa654ea67e82b46a25c
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/plugin.py
@@ -0,0 +1,88 @@
+import inspect
+import platform
+
+from .registry import PLUGIN_LAYERS
+
+if platform.system() == 'Windows':
+    import regex as re
+else:
+    import re
+
+
+def infer_abbr(class_type):
+    """Infer abbreviation from the class name.
+
+    This method will infer the abbreviation to map class types to
+    abbreviations.
+
+    Rule 1: If the class has the property "abbr", return the property.
+    Rule 2: Otherwise, the abbreviation falls back to snake case of class
+    name, e.g. the abbreviation of ``FancyBlock`` will be ``fancy_block``.
+
+    Args:
+        class_type (type): The norm layer type.
+
+    Returns:
+        str: The inferred abbreviation.
+    """
+
+    def camel2snack(word):
+        """Convert camel case word into snack case.
+
+        Modified from `inflection lib
+        <https://inflection.readthedocs.io/en/latest/#inflection.underscore>`_.
+
+        Example::
+
+            >>> camel2snack("FancyBlock")
+            'fancy_block'
+        """
+
+        word = re.sub(r'([A-Z]+)([A-Z][a-z])', r'\1_\2', word)
+        word = re.sub(r'([a-z\d])([A-Z])', r'\1_\2', word)
+        word = word.replace('-', '_')
+        return word.lower()
+
+    if not inspect.isclass(class_type):
+        raise TypeError(
+            f'class_type must be a type, but got {type(class_type)}')
+    if hasattr(class_type, '_abbr_'):
+        return class_type._abbr_
+    else:
+        return camel2snack(class_type.__name__)
+
+
+def build_plugin_layer(cfg, postfix='', **kwargs):
+    """Build plugin layer.
+
+    Args:
+        cfg (None or dict): cfg should contain:
+            type (str): identify plugin layer type.
+            layer args: args needed to instantiate a plugin layer.
+        postfix (int, str): appended into norm abbreviation to
+            create named layer. Default: ''.
+
+    Returns:
+        tuple[str, nn.Module]:
+            name (str): abbreviation + postfix
+            layer (nn.Module): created plugin layer
+    """
+    if not isinstance(cfg, dict):
+        raise TypeError('cfg must be a dict')
+    if 'type' not in cfg:
+        raise KeyError('the cfg dict must contain the key "type"')
+    cfg_ = cfg.copy()
+
+    layer_type = cfg_.pop('type')
+    if layer_type not in PLUGIN_LAYERS:
+        raise KeyError(f'Unrecognized plugin type {layer_type}')
+
+    plugin_layer = PLUGIN_LAYERS.get(layer_type)
+    abbr = infer_abbr(plugin_layer)
+
+    assert isinstance(postfix, (int, str))
+    name = abbr + str(postfix)
+
+    layer = plugin_layer(**kwargs, **cfg_)
+
+    return name, layer
diff --git a/annotator/uniformer/mmcv/cnn/bricks/registry.py b/annotator/uniformer/mmcv/cnn/bricks/registry.py
new file mode 100644
index 0000000000000000000000000000000000000000..39eabc58db4b5954478a2ac1ab91cea5e45ab055
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/registry.py
@@ -0,0 +1,16 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from annotator.uniformer.mmcv.utils import Registry
+
+CONV_LAYERS = Registry('conv layer')
+NORM_LAYERS = Registry('norm layer')
+ACTIVATION_LAYERS = Registry('activation layer')
+PADDING_LAYERS = Registry('padding layer')
+UPSAMPLE_LAYERS = Registry('upsample layer')
+PLUGIN_LAYERS = Registry('plugin layer')
+
+DROPOUT_LAYERS = Registry('drop out layers')
+POSITIONAL_ENCODING = Registry('position encoding')
+ATTENTION = Registry('attention')
+FEEDFORWARD_NETWORK = Registry('feed-forward Network')
+TRANSFORMER_LAYER = Registry('transformerLayer')
+TRANSFORMER_LAYER_SEQUENCE = Registry('transformer-layers sequence')
diff --git a/annotator/uniformer/mmcv/cnn/bricks/scale.py b/annotator/uniformer/mmcv/cnn/bricks/scale.py
new file mode 100644
index 0000000000000000000000000000000000000000..c905fffcc8bf998d18d94f927591963c428025e2
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/scale.py
@@ -0,0 +1,21 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+
+
+class Scale(nn.Module):
+    """A learnable scale parameter.
+
+    This layer scales the input by a learnable factor. It multiplies a
+    learnable scale parameter of shape (1,) with input of any shape.
+
+    Args:
+        scale (float): Initial value of scale factor. Default: 1.0
+    """
+
+    def __init__(self, scale=1.0):
+        super(Scale, self).__init__()
+        self.scale = nn.Parameter(torch.tensor(scale, dtype=torch.float))
+
+    def forward(self, x):
+        return x * self.scale
diff --git a/annotator/uniformer/mmcv/cnn/bricks/swish.py b/annotator/uniformer/mmcv/cnn/bricks/swish.py
new file mode 100644
index 0000000000000000000000000000000000000000..e2ca8ed7b749413f011ae54aac0cab27e6f0b51f
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/swish.py
@@ -0,0 +1,25 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+
+from .registry import ACTIVATION_LAYERS
+
+
+@ACTIVATION_LAYERS.register_module()
+class Swish(nn.Module):
+    """Swish Module.
+
+    This module applies the swish function:
+
+    .. math::
+        Swish(x) = x * Sigmoid(x)
+
+    Returns:
+        Tensor: The output tensor.
+    """
+
+    def __init__(self):
+        super(Swish, self).__init__()
+
+    def forward(self, x):
+        return x * torch.sigmoid(x)
diff --git a/annotator/uniformer/mmcv/cnn/bricks/transformer.py b/annotator/uniformer/mmcv/cnn/bricks/transformer.py
new file mode 100644
index 0000000000000000000000000000000000000000..e61ae0dd941a7be00b3e41a3de833ec50470a45f
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/transformer.py
@@ -0,0 +1,595 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import copy
+import warnings
+
+import torch
+import torch.nn as nn
+
+from annotator.uniformer.mmcv import ConfigDict, deprecated_api_warning
+from annotator.uniformer.mmcv.cnn import Linear, build_activation_layer, build_norm_layer
+from annotator.uniformer.mmcv.runner.base_module import BaseModule, ModuleList, Sequential
+from annotator.uniformer.mmcv.utils import build_from_cfg
+from .drop import build_dropout
+from .registry import (ATTENTION, FEEDFORWARD_NETWORK, POSITIONAL_ENCODING,
+                       TRANSFORMER_LAYER, TRANSFORMER_LAYER_SEQUENCE)
+
+# Avoid BC-breaking of importing MultiScaleDeformableAttention from this file
+try:
+    from annotator.uniformer.mmcv.ops.multi_scale_deform_attn import MultiScaleDeformableAttention  # noqa F401
+    warnings.warn(
+        ImportWarning(
+            '``MultiScaleDeformableAttention`` has been moved to '
+            '``mmcv.ops.multi_scale_deform_attn``, please change original path '  # noqa E501
+            '``from annotator.uniformer.mmcv.cnn.bricks.transformer import MultiScaleDeformableAttention`` '  # noqa E501
+            'to ``from annotator.uniformer.mmcv.ops.multi_scale_deform_attn import MultiScaleDeformableAttention`` '  # noqa E501
+        ))
+
+except ImportError:
+    warnings.warn('Fail to import ``MultiScaleDeformableAttention`` from '
+                  '``mmcv.ops.multi_scale_deform_attn``, '
+                  'You should install ``mmcv-full`` if you need this module. ')
+
+
+def build_positional_encoding(cfg, default_args=None):
+    """Builder for Position Encoding."""
+    return build_from_cfg(cfg, POSITIONAL_ENCODING, default_args)
+
+
+def build_attention(cfg, default_args=None):
+    """Builder for attention."""
+    return build_from_cfg(cfg, ATTENTION, default_args)
+
+
+def build_feedforward_network(cfg, default_args=None):
+    """Builder for feed-forward network (FFN)."""
+    return build_from_cfg(cfg, FEEDFORWARD_NETWORK, default_args)
+
+
+def build_transformer_layer(cfg, default_args=None):
+    """Builder for transformer layer."""
+    return build_from_cfg(cfg, TRANSFORMER_LAYER, default_args)
+
+
+def build_transformer_layer_sequence(cfg, default_args=None):
+    """Builder for transformer encoder and transformer decoder."""
+    return build_from_cfg(cfg, TRANSFORMER_LAYER_SEQUENCE, default_args)
+
+
+@ATTENTION.register_module()
+class MultiheadAttention(BaseModule):
+    """A wrapper for ``torch.nn.MultiheadAttention``.
+
+    This module implements MultiheadAttention with identity connection,
+    and positional encoding  is also passed as input.
+
+    Args:
+        embed_dims (int): The embedding dimension.
+        num_heads (int): Parallel attention heads.
+        attn_drop (float): A Dropout layer on attn_output_weights.
+            Default: 0.0.
+        proj_drop (float): A Dropout layer after `nn.MultiheadAttention`.
+            Default: 0.0.
+        dropout_layer (obj:`ConfigDict`): The dropout_layer used
+            when adding the shortcut.
+        init_cfg (obj:`mmcv.ConfigDict`): The Config for initialization.
+            Default: None.
+        batch_first (bool): When it is True,  Key, Query and Value are shape of
+            (batch, n, embed_dim), otherwise (n, batch, embed_dim).
+             Default to False.
+    """
+
+    def __init__(self,
+                 embed_dims,
+                 num_heads,
+                 attn_drop=0.,
+                 proj_drop=0.,
+                 dropout_layer=dict(type='Dropout', drop_prob=0.),
+                 init_cfg=None,
+                 batch_first=False,
+                 **kwargs):
+        super(MultiheadAttention, self).__init__(init_cfg)
+        if 'dropout' in kwargs:
+            warnings.warn('The arguments `dropout` in MultiheadAttention '
+                          'has been deprecated, now you can separately '
+                          'set `attn_drop`(float), proj_drop(float), '
+                          'and `dropout_layer`(dict) ')
+            attn_drop = kwargs['dropout']
+            dropout_layer['drop_prob'] = kwargs.pop('dropout')
+
+        self.embed_dims = embed_dims
+        self.num_heads = num_heads
+        self.batch_first = batch_first
+
+        self.attn = nn.MultiheadAttention(embed_dims, num_heads, attn_drop,
+                                          **kwargs)
+
+        self.proj_drop = nn.Dropout(proj_drop)
+        self.dropout_layer = build_dropout(
+            dropout_layer) if dropout_layer else nn.Identity()
+
+    @deprecated_api_warning({'residual': 'identity'},
+                            cls_name='MultiheadAttention')
+    def forward(self,
+                query,
+                key=None,
+                value=None,
+                identity=None,
+                query_pos=None,
+                key_pos=None,
+                attn_mask=None,
+                key_padding_mask=None,
+                **kwargs):
+        """Forward function for `MultiheadAttention`.
+
+        **kwargs allow passing a more general data flow when combining
+        with other operations in `transformerlayer`.
+
+        Args:
+            query (Tensor): The input query with shape [num_queries, bs,
+                embed_dims] if self.batch_first is False, else
+                [bs, num_queries embed_dims].
+            key (Tensor): The key tensor with shape [num_keys, bs,
+                embed_dims] if self.batch_first is False, else
+                [bs, num_keys, embed_dims] .
+                If None, the ``query`` will be used. Defaults to None.
+            value (Tensor): The value tensor with same shape as `key`.
+                Same in `nn.MultiheadAttention.forward`. Defaults to None.
+                If None, the `key` will be used.
+            identity (Tensor): This tensor, with the same shape as x,
+                will be used for the identity link.
+                If None, `x` will be used. Defaults to None.
+            query_pos (Tensor): The positional encoding for query, with
+                the same shape as `x`. If not None, it will
+                be added to `x` before forward function. Defaults to None.
+            key_pos (Tensor): The positional encoding for `key`, with the
+                same shape as `key`. Defaults to None. If not None, it will
+                be added to `key` before forward function. If None, and
+                `query_pos` has the same shape as `key`, then `query_pos`
+                will be used for `key_pos`. Defaults to None.
+            attn_mask (Tensor): ByteTensor mask with shape [num_queries,
+                num_keys]. Same in `nn.MultiheadAttention.forward`.
+                Defaults to None.
+            key_padding_mask (Tensor): ByteTensor with shape [bs, num_keys].
+                Defaults to None.
+
+        Returns:
+            Tensor: forwarded results with shape
+                [num_queries, bs, embed_dims]
+                if self.batch_first is False, else
+                [bs, num_queries embed_dims].
+        """
+
+        if key is None:
+            key = query
+        if value is None:
+            value = key
+        if identity is None:
+            identity = query
+        if key_pos is None:
+            if query_pos is not None:
+                # use query_pos if key_pos is not available
+                if query_pos.shape == key.shape:
+                    key_pos = query_pos
+                else:
+                    warnings.warn(f'position encoding of key is'
+                                  f'missing in {self.__class__.__name__}.')
+        if query_pos is not None:
+            query = query + query_pos
+        if key_pos is not None:
+            key = key + key_pos
+
+        # Because the dataflow('key', 'query', 'value') of
+        # ``torch.nn.MultiheadAttention`` is (num_query, batch,
+        # embed_dims), We should adjust the shape of dataflow from
+        # batch_first (batch, num_query, embed_dims) to num_query_first
+        # (num_query ,batch, embed_dims), and recover ``attn_output``
+        # from num_query_first to batch_first.
+        if self.batch_first:
+            query = query.transpose(0, 1)
+            key = key.transpose(0, 1)
+            value = value.transpose(0, 1)
+
+        out = self.attn(
+            query=query,
+            key=key,
+            value=value,
+            attn_mask=attn_mask,
+            key_padding_mask=key_padding_mask)[0]
+
+        if self.batch_first:
+            out = out.transpose(0, 1)
+
+        return identity + self.dropout_layer(self.proj_drop(out))
+
+
+@FEEDFORWARD_NETWORK.register_module()
+class FFN(BaseModule):
+    """Implements feed-forward networks (FFNs) with identity connection.
+
+    Args:
+        embed_dims (int): The feature dimension. Same as
+            `MultiheadAttention`. Defaults: 256.
+        feedforward_channels (int): The hidden dimension of FFNs.
+            Defaults: 1024.
+        num_fcs (int, optional): The number of fully-connected layers in
+            FFNs. Default: 2.
+        act_cfg (dict, optional): The activation config for FFNs.
+            Default: dict(type='ReLU')
+        ffn_drop (float, optional): Probability of an element to be
+            zeroed in FFN. Default 0.0.
+        add_identity (bool, optional): Whether to add the
+            identity connection. Default: `True`.
+        dropout_layer (obj:`ConfigDict`): The dropout_layer used
+            when adding the shortcut.
+        init_cfg (obj:`mmcv.ConfigDict`): The Config for initialization.
+            Default: None.
+    """
+
+    @deprecated_api_warning(
+        {
+            'dropout': 'ffn_drop',
+            'add_residual': 'add_identity'
+        },
+        cls_name='FFN')
+    def __init__(self,
+                 embed_dims=256,
+                 feedforward_channels=1024,
+                 num_fcs=2,
+                 act_cfg=dict(type='ReLU', inplace=True),
+                 ffn_drop=0.,
+                 dropout_layer=None,
+                 add_identity=True,
+                 init_cfg=None,
+                 **kwargs):
+        super(FFN, self).__init__(init_cfg)
+        assert num_fcs >= 2, 'num_fcs should be no less ' \
+            f'than 2. got {num_fcs}.'
+        self.embed_dims = embed_dims
+        self.feedforward_channels = feedforward_channels
+        self.num_fcs = num_fcs
+        self.act_cfg = act_cfg
+        self.activate = build_activation_layer(act_cfg)
+
+        layers = []
+        in_channels = embed_dims
+        for _ in range(num_fcs - 1):
+            layers.append(
+                Sequential(
+                    Linear(in_channels, feedforward_channels), self.activate,
+                    nn.Dropout(ffn_drop)))
+            in_channels = feedforward_channels
+        layers.append(Linear(feedforward_channels, embed_dims))
+        layers.append(nn.Dropout(ffn_drop))
+        self.layers = Sequential(*layers)
+        self.dropout_layer = build_dropout(
+            dropout_layer) if dropout_layer else torch.nn.Identity()
+        self.add_identity = add_identity
+
+    @deprecated_api_warning({'residual': 'identity'}, cls_name='FFN')
+    def forward(self, x, identity=None):
+        """Forward function for `FFN`.
+
+        The function would add x to the output tensor if residue is None.
+        """
+        out = self.layers(x)
+        if not self.add_identity:
+            return self.dropout_layer(out)
+        if identity is None:
+            identity = x
+        return identity + self.dropout_layer(out)
+
+
+@TRANSFORMER_LAYER.register_module()
+class BaseTransformerLayer(BaseModule):
+    """Base `TransformerLayer` for vision transformer.
+
+    It can be built from `mmcv.ConfigDict` and support more flexible
+    customization, for example, using any number of `FFN or LN ` and
+    use different kinds of `attention` by specifying a list of `ConfigDict`
+    named `attn_cfgs`. It is worth mentioning that it supports `prenorm`
+    when you specifying `norm` as the first element of `operation_order`.
+    More details about the `prenorm`: `On Layer Normalization in the
+    Transformer Architecture <https://arxiv.org/abs/2002.04745>`_ .
+
+    Args:
+        attn_cfgs (list[`mmcv.ConfigDict`] | obj:`mmcv.ConfigDict` | None )):
+            Configs for `self_attention` or `cross_attention` modules,
+            The order of the configs in the list should be consistent with
+            corresponding attentions in operation_order.
+            If it is a dict, all of the attention modules in operation_order
+            will be built with this config. Default: None.
+        ffn_cfgs (list[`mmcv.ConfigDict`] | obj:`mmcv.ConfigDict` | None )):
+            Configs for FFN, The order of the configs in the list should be
+            consistent with corresponding ffn in operation_order.
+            If it is a dict, all of the attention modules in operation_order
+            will be built with this config.
+        operation_order (tuple[str]): The execution order of operation
+            in transformer. Such as ('self_attn', 'norm', 'ffn', 'norm').
+            Support `prenorm` when you specifying first element as `norm`.
+            Default：None.
+        norm_cfg (dict): Config dict for normalization layer.
+            Default: dict(type='LN').
+        init_cfg (obj:`mmcv.ConfigDict`): The Config for initialization.
+            Default: None.
+        batch_first (bool): Key, Query and Value are shape
+            of (batch, n, embed_dim)
+            or (n, batch, embed_dim). Default to False.
+    """
+
+    def __init__(self,
+                 attn_cfgs=None,
+                 ffn_cfgs=dict(
+                     type='FFN',
+                     embed_dims=256,
+                     feedforward_channels=1024,
+                     num_fcs=2,
+                     ffn_drop=0.,
+                     act_cfg=dict(type='ReLU', inplace=True),
+                 ),
+                 operation_order=None,
+                 norm_cfg=dict(type='LN'),
+                 init_cfg=None,
+                 batch_first=False,
+                 **kwargs):
+
+        deprecated_args = dict(
+            feedforward_channels='feedforward_channels',
+            ffn_dropout='ffn_drop',
+            ffn_num_fcs='num_fcs')
+        for ori_name, new_name in deprecated_args.items():
+            if ori_name in kwargs:
+                warnings.warn(
+                    f'The arguments `{ori_name}` in BaseTransformerLayer '
+                    f'has been deprecated, now you should set `{new_name}` '
+                    f'and other FFN related arguments '
+                    f'to a dict named `ffn_cfgs`. ')
+                ffn_cfgs[new_name] = kwargs[ori_name]
+
+        super(BaseTransformerLayer, self).__init__(init_cfg)
+
+        self.batch_first = batch_first
+
+        assert set(operation_order) & set(
+            ['self_attn', 'norm', 'ffn', 'cross_attn']) == \
+            set(operation_order), f'The operation_order of' \
+            f' {self.__class__.__name__} should ' \
+            f'contains all four operation type ' \
+            f"{['self_attn', 'norm', 'ffn', 'cross_attn']}"
+
+        num_attn = operation_order.count('self_attn') + operation_order.count(
+            'cross_attn')
+        if isinstance(attn_cfgs, dict):
+            attn_cfgs = [copy.deepcopy(attn_cfgs) for _ in range(num_attn)]
+        else:
+            assert num_attn == len(attn_cfgs), f'The length ' \
+                f'of attn_cfg {num_attn} is ' \
+                f'not consistent with the number of attention' \
+                f'in operation_order {operation_order}.'
+
+        self.num_attn = num_attn
+        self.operation_order = operation_order
+        self.norm_cfg = norm_cfg
+        self.pre_norm = operation_order[0] == 'norm'
+        self.attentions = ModuleList()
+
+        index = 0
+        for operation_name in operation_order:
+            if operation_name in ['self_attn', 'cross_attn']:
+                if 'batch_first' in attn_cfgs[index]:
+                    assert self.batch_first == attn_cfgs[index]['batch_first']
+                else:
+                    attn_cfgs[index]['batch_first'] = self.batch_first
+                attention = build_attention(attn_cfgs[index])
+                # Some custom attentions used as `self_attn`
+                # or `cross_attn` can have different behavior.
+                attention.operation_name = operation_name
+                self.attentions.append(attention)
+                index += 1
+
+        self.embed_dims = self.attentions[0].embed_dims
+
+        self.ffns = ModuleList()
+        num_ffns = operation_order.count('ffn')
+        if isinstance(ffn_cfgs, dict):
+            ffn_cfgs = ConfigDict(ffn_cfgs)
+        if isinstance(ffn_cfgs, dict):
+            ffn_cfgs = [copy.deepcopy(ffn_cfgs) for _ in range(num_ffns)]
+        assert len(ffn_cfgs) == num_ffns
+        for ffn_index in range(num_ffns):
+            if 'embed_dims' not in ffn_cfgs[ffn_index]:
+                ffn_cfgs['embed_dims'] = self.embed_dims
+            else:
+                assert ffn_cfgs[ffn_index]['embed_dims'] == self.embed_dims
+            self.ffns.append(
+                build_feedforward_network(ffn_cfgs[ffn_index],
+                                          dict(type='FFN')))
+
+        self.norms = ModuleList()
+        num_norms = operation_order.count('norm')
+        for _ in range(num_norms):
+            self.norms.append(build_norm_layer(norm_cfg, self.embed_dims)[1])
+
+    def forward(self,
+                query,
+                key=None,
+                value=None,
+                query_pos=None,
+                key_pos=None,
+                attn_masks=None,
+                query_key_padding_mask=None,
+                key_padding_mask=None,
+                **kwargs):
+        """Forward function for `TransformerDecoderLayer`.
+
+        **kwargs contains some specific arguments of attentions.
+
+        Args:
+            query (Tensor): The input query with shape
+                [num_queries, bs, embed_dims] if
+                self.batch_first is False, else
+                [bs, num_queries embed_dims].
+            key (Tensor): The key tensor with shape [num_keys, bs,
+                embed_dims] if self.batch_first is False, else
+                [bs, num_keys, embed_dims] .
+            value (Tensor): The value tensor with same shape as `key`.
+            query_pos (Tensor): The positional encoding for `query`.
+                Default: None.
+            key_pos (Tensor): The positional encoding for `key`.
+                Default: None.
+            attn_masks (List[Tensor] | None): 2D Tensor used in
+                calculation of corresponding attention. The length of
+                it should equal to the number of `attention` in
+                `operation_order`. Default: None.
+            query_key_padding_mask (Tensor): ByteTensor for `query`, with
+                shape [bs, num_queries]. Only used in `self_attn` layer.
+                Defaults to None.
+            key_padding_mask (Tensor): ByteTensor for `query`, with
+                shape [bs, num_keys]. Default: None.
+
+        Returns:
+            Tensor: forwarded results with shape [num_queries, bs, embed_dims].
+        """
+
+        norm_index = 0
+        attn_index = 0
+        ffn_index = 0
+        identity = query
+        if attn_masks is None:
+            attn_masks = [None for _ in range(self.num_attn)]
+        elif isinstance(attn_masks, torch.Tensor):
+            attn_masks = [
+                copy.deepcopy(attn_masks) for _ in range(self.num_attn)
+            ]
+            warnings.warn(f'Use same attn_mask in all attentions in '
+                          f'{self.__class__.__name__} ')
+        else:
+            assert len(attn_masks) == self.num_attn, f'The length of ' \
+                        f'attn_masks {len(attn_masks)} must be equal ' \
+                        f'to the number of attention in ' \
+                        f'operation_order {self.num_attn}'
+
+        for layer in self.operation_order:
+            if layer == 'self_attn':
+                temp_key = temp_value = query
+                query = self.attentions[attn_index](
+                    query,
+                    temp_key,
+                    temp_value,
+                    identity if self.pre_norm else None,
+                    query_pos=query_pos,
+                    key_pos=query_pos,
+                    attn_mask=attn_masks[attn_index],
+                    key_padding_mask=query_key_padding_mask,
+                    **kwargs)
+                attn_index += 1
+                identity = query
+
+            elif layer == 'norm':
+                query = self.norms[norm_index](query)
+                norm_index += 1
+
+            elif layer == 'cross_attn':
+                query = self.attentions[attn_index](
+                    query,
+                    key,
+                    value,
+                    identity if self.pre_norm else None,
+                    query_pos=query_pos,
+                    key_pos=key_pos,
+                    attn_mask=attn_masks[attn_index],
+                    key_padding_mask=key_padding_mask,
+                    **kwargs)
+                attn_index += 1
+                identity = query
+
+            elif layer == 'ffn':
+                query = self.ffns[ffn_index](
+                    query, identity if self.pre_norm else None)
+                ffn_index += 1
+
+        return query
+
+
+@TRANSFORMER_LAYER_SEQUENCE.register_module()
+class TransformerLayerSequence(BaseModule):
+    """Base class for TransformerEncoder and TransformerDecoder in vision
+    transformer.
+
+    As base-class of Encoder and Decoder in vision transformer.
+    Support customization such as specifying different kind
+    of `transformer_layer` in `transformer_coder`.
+
+    Args:
+        transformerlayer (list[obj:`mmcv.ConfigDict`] |
+            obj:`mmcv.ConfigDict`): Config of transformerlayer
+            in TransformerCoder. If it is obj:`mmcv.ConfigDict`,
+             it would be repeated `num_layer` times to a
+             list[`mmcv.ConfigDict`]. Default: None.
+        num_layers (int): The number of `TransformerLayer`. Default: None.
+        init_cfg (obj:`mmcv.ConfigDict`): The Config for initialization.
+            Default: None.
+    """
+
+    def __init__(self, transformerlayers=None, num_layers=None, init_cfg=None):
+        super(TransformerLayerSequence, self).__init__(init_cfg)
+        if isinstance(transformerlayers, dict):
+            transformerlayers = [
+                copy.deepcopy(transformerlayers) for _ in range(num_layers)
+            ]
+        else:
+            assert isinstance(transformerlayers, list) and \
+                   len(transformerlayers) == num_layers
+        self.num_layers = num_layers
+        self.layers = ModuleList()
+        for i in range(num_layers):
+            self.layers.append(build_transformer_layer(transformerlayers[i]))
+        self.embed_dims = self.layers[0].embed_dims
+        self.pre_norm = self.layers[0].pre_norm
+
+    def forward(self,
+                query,
+                key,
+                value,
+                query_pos=None,
+                key_pos=None,
+                attn_masks=None,
+                query_key_padding_mask=None,
+                key_padding_mask=None,
+                **kwargs):
+        """Forward function for `TransformerCoder`.
+
+        Args:
+            query (Tensor): Input query with shape
+                `(num_queries, bs, embed_dims)`.
+            key (Tensor): The key tensor with shape
+                `(num_keys, bs, embed_dims)`.
+            value (Tensor): The value tensor with shape
+                `(num_keys, bs, embed_dims)`.
+            query_pos (Tensor): The positional encoding for `query`.
+                Default: None.
+            key_pos (Tensor): The positional encoding for `key`.
+                Default: None.
+            attn_masks (List[Tensor], optional): Each element is 2D Tensor
+                which is used in calculation of corresponding attention in
+                operation_order. Default: None.
+            query_key_padding_mask (Tensor): ByteTensor for `query`, with
+                shape [bs, num_queries]. Only used in self-attention
+                Default: None.
+            key_padding_mask (Tensor): ByteTensor for `query`, with
+                shape [bs, num_keys]. Default: None.
+
+        Returns:
+            Tensor:  results with shape [num_queries, bs, embed_dims].
+        """
+        for layer in self.layers:
+            query = layer(
+                query,
+                key,
+                value,
+                query_pos=query_pos,
+                key_pos=key_pos,
+                attn_masks=attn_masks,
+                query_key_padding_mask=query_key_padding_mask,
+                key_padding_mask=key_padding_mask,
+                **kwargs)
+        return query
diff --git a/annotator/uniformer/mmcv/cnn/bricks/upsample.py b/annotator/uniformer/mmcv/cnn/bricks/upsample.py
new file mode 100644
index 0000000000000000000000000000000000000000..a1a353767d0ce8518f0d7289bed10dba0178ed12
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/upsample.py
@@ -0,0 +1,84 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch.nn as nn
+import torch.nn.functional as F
+
+from ..utils import xavier_init
+from .registry import UPSAMPLE_LAYERS
+
+UPSAMPLE_LAYERS.register_module('nearest', module=nn.Upsample)
+UPSAMPLE_LAYERS.register_module('bilinear', module=nn.Upsample)
+
+
+@UPSAMPLE_LAYERS.register_module(name='pixel_shuffle')
+class PixelShufflePack(nn.Module):
+    """Pixel Shuffle upsample layer.
+
+    This module packs `F.pixel_shuffle()` and a nn.Conv2d module together to
+    achieve a simple upsampling with pixel shuffle.
+
+    Args:
+        in_channels (int): Number of input channels.
+        out_channels (int): Number of output channels.
+        scale_factor (int): Upsample ratio.
+        upsample_kernel (int): Kernel size of the conv layer to expand the
+            channels.
+    """
+
+    def __init__(self, in_channels, out_channels, scale_factor,
+                 upsample_kernel):
+        super(PixelShufflePack, self).__init__()
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.scale_factor = scale_factor
+        self.upsample_kernel = upsample_kernel
+        self.upsample_conv = nn.Conv2d(
+            self.in_channels,
+            self.out_channels * scale_factor * scale_factor,
+            self.upsample_kernel,
+            padding=(self.upsample_kernel - 1) // 2)
+        self.init_weights()
+
+    def init_weights(self):
+        xavier_init(self.upsample_conv, distribution='uniform')
+
+    def forward(self, x):
+        x = self.upsample_conv(x)
+        x = F.pixel_shuffle(x, self.scale_factor)
+        return x
+
+
+def build_upsample_layer(cfg, *args, **kwargs):
+    """Build upsample layer.
+
+    Args:
+        cfg (dict): The upsample layer config, which should contain:
+
+            - type (str): Layer type.
+            - scale_factor (int): Upsample ratio, which is not applicable to
+                deconv.
+            - layer args: Args needed to instantiate a upsample layer.
+        args (argument list): Arguments passed to the ``__init__``
+            method of the corresponding conv layer.
+        kwargs (keyword arguments): Keyword arguments passed to the
+            ``__init__`` method of the corresponding conv layer.
+
+    Returns:
+        nn.Module: Created upsample layer.
+    """
+    if not isinstance(cfg, dict):
+        raise TypeError(f'cfg must be a dict, but got {type(cfg)}')
+    if 'type' not in cfg:
+        raise KeyError(
+            f'the cfg dict must contain the key "type", but got {cfg}')
+    cfg_ = cfg.copy()
+
+    layer_type = cfg_.pop('type')
+    if layer_type not in UPSAMPLE_LAYERS:
+        raise KeyError(f'Unrecognized upsample type {layer_type}')
+    else:
+        upsample = UPSAMPLE_LAYERS.get(layer_type)
+
+    if upsample is nn.Upsample:
+        cfg_['mode'] = layer_type
+    layer = upsample(*args, **kwargs, **cfg_)
+    return layer
diff --git a/annotator/uniformer/mmcv/cnn/bricks/wrappers.py b/annotator/uniformer/mmcv/cnn/bricks/wrappers.py
new file mode 100644
index 0000000000000000000000000000000000000000..8aebf67bf52355a513f21756ee74fe510902d075
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/bricks/wrappers.py
@@ -0,0 +1,180 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+r"""Modified from https://github.com/facebookresearch/detectron2/blob/master/detectron2/layers/wrappers.py  # noqa: E501
+
+Wrap some nn modules to support empty tensor input. Currently, these wrappers
+are mainly used in mask heads like fcn_mask_head and maskiou_heads since mask
+heads are trained on only positive RoIs.
+"""
+import math
+
+import torch
+import torch.nn as nn
+from torch.nn.modules.utils import _pair, _triple
+
+from .registry import CONV_LAYERS, UPSAMPLE_LAYERS
+
+if torch.__version__ == 'parrots':
+    TORCH_VERSION = torch.__version__
+else:
+    # torch.__version__ could be 1.3.1+cu92, we only need the first two
+    # for comparison
+    TORCH_VERSION = tuple(int(x) for x in torch.__version__.split('.')[:2])
+
+
+def obsolete_torch_version(torch_version, version_threshold):
+    return torch_version == 'parrots' or torch_version <= version_threshold
+
+
+class NewEmptyTensorOp(torch.autograd.Function):
+
+    @staticmethod
+    def forward(ctx, x, new_shape):
+        ctx.shape = x.shape
+        return x.new_empty(new_shape)
+
+    @staticmethod
+    def backward(ctx, grad):
+        shape = ctx.shape
+        return NewEmptyTensorOp.apply(grad, shape), None
+
+
+@CONV_LAYERS.register_module('Conv', force=True)
+class Conv2d(nn.Conv2d):
+
+    def forward(self, x):
+        if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 4)):
+            out_shape = [x.shape[0], self.out_channels]
+            for i, k, p, s, d in zip(x.shape[-2:], self.kernel_size,
+                                     self.padding, self.stride, self.dilation):
+                o = (i + 2 * p - (d * (k - 1) + 1)) // s + 1
+                out_shape.append(o)
+            empty = NewEmptyTensorOp.apply(x, out_shape)
+            if self.training:
+                # produce dummy gradient to avoid DDP warning.
+                dummy = sum(x.view(-1)[0] for x in self.parameters()) * 0.0
+                return empty + dummy
+            else:
+                return empty
+
+        return super().forward(x)
+
+
+@CONV_LAYERS.register_module('Conv3d', force=True)
+class Conv3d(nn.Conv3d):
+
+    def forward(self, x):
+        if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 4)):
+            out_shape = [x.shape[0], self.out_channels]
+            for i, k, p, s, d in zip(x.shape[-3:], self.kernel_size,
+                                     self.padding, self.stride, self.dilation):
+                o = (i + 2 * p - (d * (k - 1) + 1)) // s + 1
+                out_shape.append(o)
+            empty = NewEmptyTensorOp.apply(x, out_shape)
+            if self.training:
+                # produce dummy gradient to avoid DDP warning.
+                dummy = sum(x.view(-1)[0] for x in self.parameters()) * 0.0
+                return empty + dummy
+            else:
+                return empty
+
+        return super().forward(x)
+
+
+@CONV_LAYERS.register_module()
+@CONV_LAYERS.register_module('deconv')
+@UPSAMPLE_LAYERS.register_module('deconv', force=True)
+class ConvTranspose2d(nn.ConvTranspose2d):
+
+    def forward(self, x):
+        if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 4)):
+            out_shape = [x.shape[0], self.out_channels]
+            for i, k, p, s, d, op in zip(x.shape[-2:], self.kernel_size,
+                                         self.padding, self.stride,
+                                         self.dilation, self.output_padding):
+                out_shape.append((i - 1) * s - 2 * p + (d * (k - 1) + 1) + op)
+            empty = NewEmptyTensorOp.apply(x, out_shape)
+            if self.training:
+                # produce dummy gradient to avoid DDP warning.
+                dummy = sum(x.view(-1)[0] for x in self.parameters()) * 0.0
+                return empty + dummy
+            else:
+                return empty
+
+        return super().forward(x)
+
+
+@CONV_LAYERS.register_module()
+@CONV_LAYERS.register_module('deconv3d')
+@UPSAMPLE_LAYERS.register_module('deconv3d', force=True)
+class ConvTranspose3d(nn.ConvTranspose3d):
+
+    def forward(self, x):
+        if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 4)):
+            out_shape = [x.shape[0], self.out_channels]
+            for i, k, p, s, d, op in zip(x.shape[-3:], self.kernel_size,
+                                         self.padding, self.stride,
+                                         self.dilation, self.output_padding):
+                out_shape.append((i - 1) * s - 2 * p + (d * (k - 1) + 1) + op)
+            empty = NewEmptyTensorOp.apply(x, out_shape)
+            if self.training:
+                # produce dummy gradient to avoid DDP warning.
+                dummy = sum(x.view(-1)[0] for x in self.parameters()) * 0.0
+                return empty + dummy
+            else:
+                return empty
+
+        return super().forward(x)
+
+
+class MaxPool2d(nn.MaxPool2d):
+
+    def forward(self, x):
+        # PyTorch 1.9 does not support empty tensor inference yet
+        if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 9)):
+            out_shape = list(x.shape[:2])
+            for i, k, p, s, d in zip(x.shape[-2:], _pair(self.kernel_size),
+                                     _pair(self.padding), _pair(self.stride),
+                                     _pair(self.dilation)):
+                o = (i + 2 * p - (d * (k - 1) + 1)) / s + 1
+                o = math.ceil(o) if self.ceil_mode else math.floor(o)
+                out_shape.append(o)
+            empty = NewEmptyTensorOp.apply(x, out_shape)
+            return empty
+
+        return super().forward(x)
+
+
+class MaxPool3d(nn.MaxPool3d):
+
+    def forward(self, x):
+        # PyTorch 1.9 does not support empty tensor inference yet
+        if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 9)):
+            out_shape = list(x.shape[:2])
+            for i, k, p, s, d in zip(x.shape[-3:], _triple(self.kernel_size),
+                                     _triple(self.padding),
+                                     _triple(self.stride),
+                                     _triple(self.dilation)):
+                o = (i + 2 * p - (d * (k - 1) + 1)) / s + 1
+                o = math.ceil(o) if self.ceil_mode else math.floor(o)
+                out_shape.append(o)
+            empty = NewEmptyTensorOp.apply(x, out_shape)
+            return empty
+
+        return super().forward(x)
+
+
+class Linear(torch.nn.Linear):
+
+    def forward(self, x):
+        # empty tensor forward of Linear layer is supported in Pytorch 1.6
+        if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 5)):
+            out_shape = [x.shape[0], self.out_features]
+            empty = NewEmptyTensorOp.apply(x, out_shape)
+            if self.training:
+                # produce dummy gradient to avoid DDP warning.
+                dummy = sum(x.view(-1)[0] for x in self.parameters()) * 0.0
+                return empty + dummy
+            else:
+                return empty
+
+        return super().forward(x)
diff --git a/annotator/uniformer/mmcv/cnn/builder.py b/annotator/uniformer/mmcv/cnn/builder.py
new file mode 100644
index 0000000000000000000000000000000000000000..7567316c566bd3aca6d8f65a84b00e9e890948a7
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/builder.py
@@ -0,0 +1,30 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from ..runner import Sequential
+from ..utils import Registry, build_from_cfg
+
+
+def build_model_from_cfg(cfg, registry, default_args=None):
+    """Build a PyTorch model from config dict(s). Different from
+    ``build_from_cfg``, if cfg is a list, a ``nn.Sequential`` will be built.
+
+    Args:
+        cfg (dict, list[dict]): The config of modules, is is either a config
+            dict or a list of config dicts. If cfg is a list, a
+            the built modules will be wrapped with ``nn.Sequential``.
+        registry (:obj:`Registry`): A registry the module belongs to.
+        default_args (dict, optional): Default arguments to build the module.
+            Defaults to None.
+
+    Returns:
+        nn.Module: A built nn module.
+    """
+    if isinstance(cfg, list):
+        modules = [
+            build_from_cfg(cfg_, registry, default_args) for cfg_ in cfg
+        ]
+        return Sequential(*modules)
+    else:
+        return build_from_cfg(cfg, registry, default_args)
+
+
+MODELS = Registry('model', build_func=build_model_from_cfg)
diff --git a/annotator/uniformer/mmcv/cnn/resnet.py b/annotator/uniformer/mmcv/cnn/resnet.py
new file mode 100644
index 0000000000000000000000000000000000000000..1cb3ac057ee2d52c46fc94685b5d4e698aad8d5f
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/resnet.py
@@ -0,0 +1,316 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import logging
+
+import torch.nn as nn
+import torch.utils.checkpoint as cp
+
+from .utils import constant_init, kaiming_init
+
+
+def conv3x3(in_planes, out_planes, stride=1, dilation=1):
+    """3x3 convolution with padding."""
+    return nn.Conv2d(
+        in_planes,
+        out_planes,
+        kernel_size=3,
+        stride=stride,
+        padding=dilation,
+        dilation=dilation,
+        bias=False)
+
+
+class BasicBlock(nn.Module):
+    expansion = 1
+
+    def __init__(self,
+                 inplanes,
+                 planes,
+                 stride=1,
+                 dilation=1,
+                 downsample=None,
+                 style='pytorch',
+                 with_cp=False):
+        super(BasicBlock, self).__init__()
+        assert style in ['pytorch', 'caffe']
+        self.conv1 = conv3x3(inplanes, planes, stride, dilation)
+        self.bn1 = nn.BatchNorm2d(planes)
+        self.relu = nn.ReLU(inplace=True)
+        self.conv2 = conv3x3(planes, planes)
+        self.bn2 = nn.BatchNorm2d(planes)
+        self.downsample = downsample
+        self.stride = stride
+        self.dilation = dilation
+        assert not with_cp
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+
+        out = self.conv2(out)
+        out = self.bn2(out)
+
+        if self.downsample is not None:
+            residual = self.downsample(x)
+
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class Bottleneck(nn.Module):
+    expansion = 4
+
+    def __init__(self,
+                 inplanes,
+                 planes,
+                 stride=1,
+                 dilation=1,
+                 downsample=None,
+                 style='pytorch',
+                 with_cp=False):
+        """Bottleneck block.
+
+        If style is "pytorch", the stride-two layer is the 3x3 conv layer, if
+        it is "caffe", the stride-two layer is the first 1x1 conv layer.
+        """
+        super(Bottleneck, self).__init__()
+        assert style in ['pytorch', 'caffe']
+        if style == 'pytorch':
+            conv1_stride = 1
+            conv2_stride = stride
+        else:
+            conv1_stride = stride
+            conv2_stride = 1
+        self.conv1 = nn.Conv2d(
+            inplanes, planes, kernel_size=1, stride=conv1_stride, bias=False)
+        self.conv2 = nn.Conv2d(
+            planes,
+            planes,
+            kernel_size=3,
+            stride=conv2_stride,
+            padding=dilation,
+            dilation=dilation,
+            bias=False)
+
+        self.bn1 = nn.BatchNorm2d(planes)
+        self.bn2 = nn.BatchNorm2d(planes)
+        self.conv3 = nn.Conv2d(
+            planes, planes * self.expansion, kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.relu = nn.ReLU(inplace=True)
+        self.downsample = downsample
+        self.stride = stride
+        self.dilation = dilation
+        self.with_cp = with_cp
+
+    def forward(self, x):
+
+        def _inner_forward(x):
+            residual = x
+
+            out = self.conv1(x)
+            out = self.bn1(out)
+            out = self.relu(out)
+
+            out = self.conv2(out)
+            out = self.bn2(out)
+            out = self.relu(out)
+
+            out = self.conv3(out)
+            out = self.bn3(out)
+
+            if self.downsample is not None:
+                residual = self.downsample(x)
+
+            out += residual
+
+            return out
+
+        if self.with_cp and x.requires_grad:
+            out = cp.checkpoint(_inner_forward, x)
+        else:
+            out = _inner_forward(x)
+
+        out = self.relu(out)
+
+        return out
+
+
+def make_res_layer(block,
+                   inplanes,
+                   planes,
+                   blocks,
+                   stride=1,
+                   dilation=1,
+                   style='pytorch',
+                   with_cp=False):
+    downsample = None
+    if stride != 1 or inplanes != planes * block.expansion:
+        downsample = nn.Sequential(
+            nn.Conv2d(
+                inplanes,
+                planes * block.expansion,
+                kernel_size=1,
+                stride=stride,
+                bias=False),
+            nn.BatchNorm2d(planes * block.expansion),
+        )
+
+    layers = []
+    layers.append(
+        block(
+            inplanes,
+            planes,
+            stride,
+            dilation,
+            downsample,
+            style=style,
+            with_cp=with_cp))
+    inplanes = planes * block.expansion
+    for _ in range(1, blocks):
+        layers.append(
+            block(inplanes, planes, 1, dilation, style=style, with_cp=with_cp))
+
+    return nn.Sequential(*layers)
+
+
+class ResNet(nn.Module):
+    """ResNet backbone.
+
+    Args:
+        depth (int): Depth of resnet, from {18, 34, 50, 101, 152}.
+        num_stages (int): Resnet stages, normally 4.
+        strides (Sequence[int]): Strides of the first block of each stage.
+        dilations (Sequence[int]): Dilation of each stage.
+        out_indices (Sequence[int]): Output from which stages.
+        style (str): `pytorch` or `caffe`. If set to "pytorch", the stride-two
+            layer is the 3x3 conv layer, otherwise the stride-two layer is
+            the first 1x1 conv layer.
+        frozen_stages (int): Stages to be frozen (all param fixed). -1 means
+            not freezing any parameters.
+        bn_eval (bool): Whether to set BN layers as eval mode, namely, freeze
+            running stats (mean and var).
+        bn_frozen (bool): Whether to freeze weight and bias of BN layers.
+        with_cp (bool): Use checkpoint or not. Using checkpoint will save some
+            memory while slowing down the training speed.
+    """
+
+    arch_settings = {
+        18: (BasicBlock, (2, 2, 2, 2)),
+        34: (BasicBlock, (3, 4, 6, 3)),
+        50: (Bottleneck, (3, 4, 6, 3)),
+        101: (Bottleneck, (3, 4, 23, 3)),
+        152: (Bottleneck, (3, 8, 36, 3))
+    }
+
+    def __init__(self,
+                 depth,
+                 num_stages=4,
+                 strides=(1, 2, 2, 2),
+                 dilations=(1, 1, 1, 1),
+                 out_indices=(0, 1, 2, 3),
+                 style='pytorch',
+                 frozen_stages=-1,
+                 bn_eval=True,
+                 bn_frozen=False,
+                 with_cp=False):
+        super(ResNet, self).__init__()
+        if depth not in self.arch_settings:
+            raise KeyError(f'invalid depth {depth} for resnet')
+        assert num_stages >= 1 and num_stages <= 4
+        block, stage_blocks = self.arch_settings[depth]
+        stage_blocks = stage_blocks[:num_stages]
+        assert len(strides) == len(dilations) == num_stages
+        assert max(out_indices) < num_stages
+
+        self.out_indices = out_indices
+        self.style = style
+        self.frozen_stages = frozen_stages
+        self.bn_eval = bn_eval
+        self.bn_frozen = bn_frozen
+        self.with_cp = with_cp
+
+        self.inplanes = 64
+        self.conv1 = nn.Conv2d(
+            3, 64, kernel_size=7, stride=2, padding=3, bias=False)
+        self.bn1 = nn.BatchNorm2d(64)
+        self.relu = nn.ReLU(inplace=True)
+        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
+
+        self.res_layers = []
+        for i, num_blocks in enumerate(stage_blocks):
+            stride = strides[i]
+            dilation = dilations[i]
+            planes = 64 * 2**i
+            res_layer = make_res_layer(
+                block,
+                self.inplanes,
+                planes,
+                num_blocks,
+                stride=stride,
+                dilation=dilation,
+                style=self.style,
+                with_cp=with_cp)
+            self.inplanes = planes * block.expansion
+            layer_name = f'layer{i + 1}'
+            self.add_module(layer_name, res_layer)
+            self.res_layers.append(layer_name)
+
+        self.feat_dim = block.expansion * 64 * 2**(len(stage_blocks) - 1)
+
+    def init_weights(self, pretrained=None):
+        if isinstance(pretrained, str):
+            logger = logging.getLogger()
+            from ..runner import load_checkpoint
+            load_checkpoint(self, pretrained, strict=False, logger=logger)
+        elif pretrained is None:
+            for m in self.modules():
+                if isinstance(m, nn.Conv2d):
+                    kaiming_init(m)
+                elif isinstance(m, nn.BatchNorm2d):
+                    constant_init(m, 1)
+        else:
+            raise TypeError('pretrained must be a str or None')
+
+    def forward(self, x):
+        x = self.conv1(x)
+        x = self.bn1(x)
+        x = self.relu(x)
+        x = self.maxpool(x)
+        outs = []
+        for i, layer_name in enumerate(self.res_layers):
+            res_layer = getattr(self, layer_name)
+            x = res_layer(x)
+            if i in self.out_indices:
+                outs.append(x)
+        if len(outs) == 1:
+            return outs[0]
+        else:
+            return tuple(outs)
+
+    def train(self, mode=True):
+        super(ResNet, self).train(mode)
+        if self.bn_eval:
+            for m in self.modules():
+                if isinstance(m, nn.BatchNorm2d):
+                    m.eval()
+                    if self.bn_frozen:
+                        for params in m.parameters():
+                            params.requires_grad = False
+        if mode and self.frozen_stages >= 0:
+            for param in self.conv1.parameters():
+                param.requires_grad = False
+            for param in self.bn1.parameters():
+                param.requires_grad = False
+            self.bn1.eval()
+            self.bn1.weight.requires_grad = False
+            self.bn1.bias.requires_grad = False
+            for i in range(1, self.frozen_stages + 1):
+                mod = getattr(self, f'layer{i}')
+                mod.eval()
+                for param in mod.parameters():
+                    param.requires_grad = False
diff --git a/annotator/uniformer/mmcv/cnn/utils/__init__.py b/annotator/uniformer/mmcv/cnn/utils/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..a263e31c1e3977712827ca229bbc04910b4e928e
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/utils/__init__.py
@@ -0,0 +1,19 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .flops_counter import get_model_complexity_info
+from .fuse_conv_bn import fuse_conv_bn
+from .sync_bn import revert_sync_batchnorm
+from .weight_init import (INITIALIZERS, Caffe2XavierInit, ConstantInit,
+                          KaimingInit, NormalInit, PretrainedInit,
+                          TruncNormalInit, UniformInit, XavierInit,
+                          bias_init_with_prob, caffe2_xavier_init,
+                          constant_init, initialize, kaiming_init, normal_init,
+                          trunc_normal_init, uniform_init, xavier_init)
+
+__all__ = [
+    'get_model_complexity_info', 'bias_init_with_prob', 'caffe2_xavier_init',
+    'constant_init', 'kaiming_init', 'normal_init', 'trunc_normal_init',
+    'uniform_init', 'xavier_init', 'fuse_conv_bn', 'initialize',
+    'INITIALIZERS', 'ConstantInit', 'XavierInit', 'NormalInit',
+    'TruncNormalInit', 'UniformInit', 'KaimingInit', 'PretrainedInit',
+    'Caffe2XavierInit', 'revert_sync_batchnorm'
+]
diff --git a/annotator/uniformer/mmcv/cnn/utils/flops_counter.py b/annotator/uniformer/mmcv/cnn/utils/flops_counter.py
new file mode 100644
index 0000000000000000000000000000000000000000..d10af5feca7f4b8c0ba359b7b1c826f754e048be
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/utils/flops_counter.py
@@ -0,0 +1,599 @@
+# Modified from flops-counter.pytorch by Vladislav Sovrasov
+# original repo: https://github.com/sovrasov/flops-counter.pytorch
+
+# MIT License
+
+# Copyright (c) 2018 Vladislav Sovrasov
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+import sys
+from functools import partial
+
+import numpy as np
+import torch
+import torch.nn as nn
+
+import annotator.uniformer.mmcv as mmcv
+
+
+def get_model_complexity_info(model,
+                              input_shape,
+                              print_per_layer_stat=True,
+                              as_strings=True,
+                              input_constructor=None,
+                              flush=False,
+                              ost=sys.stdout):
+    """Get complexity information of a model.
+
+    This method can calculate FLOPs and parameter counts of a model with
+    corresponding input shape. It can also print complexity information for
+    each layer in a model.
+
+    Supported layers are listed as below:
+        - Convolutions: ``nn.Conv1d``, ``nn.Conv2d``, ``nn.Conv3d``.
+        - Activations: ``nn.ReLU``, ``nn.PReLU``, ``nn.ELU``, ``nn.LeakyReLU``,
+            ``nn.ReLU6``.
+        - Poolings: ``nn.MaxPool1d``, ``nn.MaxPool2d``, ``nn.MaxPool3d``,
+            ``nn.AvgPool1d``, ``nn.AvgPool2d``, ``nn.AvgPool3d``,
+            ``nn.AdaptiveMaxPool1d``, ``nn.AdaptiveMaxPool2d``,
+            ``nn.AdaptiveMaxPool3d``, ``nn.AdaptiveAvgPool1d``,
+            ``nn.AdaptiveAvgPool2d``, ``nn.AdaptiveAvgPool3d``.
+        - BatchNorms: ``nn.BatchNorm1d``, ``nn.BatchNorm2d``,
+            ``nn.BatchNorm3d``, ``nn.GroupNorm``, ``nn.InstanceNorm1d``,
+            ``InstanceNorm2d``, ``InstanceNorm3d``, ``nn.LayerNorm``.
+        - Linear: ``nn.Linear``.
+        - Deconvolution: ``nn.ConvTranspose2d``.
+        - Upsample: ``nn.Upsample``.
+
+    Args:
+        model (nn.Module): The model for complexity calculation.
+        input_shape (tuple): Input shape used for calculation.
+        print_per_layer_stat (bool): Whether to print complexity information
+            for each layer in a model. Default: True.
+        as_strings (bool): Output FLOPs and params counts in a string form.
+            Default: True.
+        input_constructor (None | callable): If specified, it takes a callable
+            method that generates input. otherwise, it will generate a random
+            tensor with input shape to calculate FLOPs. Default: None.
+        flush (bool): same as that in :func:`print`. Default: False.
+        ost (stream): same as ``file`` param in :func:`print`.
+            Default: sys.stdout.
+
+    Returns:
+        tuple[float | str]: If ``as_strings`` is set to True, it will return
+            FLOPs and parameter counts in a string format. otherwise, it will
+            return those in a float number format.
+    """
+    assert type(input_shape) is tuple
+    assert len(input_shape) >= 1
+    assert isinstance(model, nn.Module)
+    flops_model = add_flops_counting_methods(model)
+    flops_model.eval()
+    flops_model.start_flops_count()
+    if input_constructor:
+        input = input_constructor(input_shape)
+        _ = flops_model(**input)
+    else:
+        try:
+            batch = torch.ones(()).new_empty(
+                (1, *input_shape),
+                dtype=next(flops_model.parameters()).dtype,
+                device=next(flops_model.parameters()).device)
+        except StopIteration:
+            # Avoid StopIteration for models which have no parameters,
+            # like `nn.Relu()`, `nn.AvgPool2d`, etc.
+            batch = torch.ones(()).new_empty((1, *input_shape))
+
+        _ = flops_model(batch)
+
+    flops_count, params_count = flops_model.compute_average_flops_cost()
+    if print_per_layer_stat:
+        print_model_with_flops(
+            flops_model, flops_count, params_count, ost=ost, flush=flush)
+    flops_model.stop_flops_count()
+
+    if as_strings:
+        return flops_to_string(flops_count), params_to_string(params_count)
+
+    return flops_count, params_count
+
+
+def flops_to_string(flops, units='GFLOPs', precision=2):
+    """Convert FLOPs number into a string.
+
+    Note that Here we take a multiply-add counts as one FLOP.
+
+    Args:
+        flops (float): FLOPs number to be converted.
+        units (str | None): Converted FLOPs units. Options are None, 'GFLOPs',
+            'MFLOPs', 'KFLOPs', 'FLOPs'. If set to None, it will automatically
+            choose the most suitable unit for FLOPs. Default: 'GFLOPs'.
+        precision (int): Digit number after the decimal point. Default: 2.
+
+    Returns:
+        str: The converted FLOPs number with units.
+
+    Examples:
+        >>> flops_to_string(1e9)
+        '1.0 GFLOPs'
+        >>> flops_to_string(2e5, 'MFLOPs')
+        '0.2 MFLOPs'
+        >>> flops_to_string(3e-9, None)
+        '3e-09 FLOPs'
+    """
+    if units is None:
+        if flops // 10**9 > 0:
+            return str(round(flops / 10.**9, precision)) + ' GFLOPs'
+        elif flops // 10**6 > 0:
+            return str(round(flops / 10.**6, precision)) + ' MFLOPs'
+        elif flops // 10**3 > 0:
+            return str(round(flops / 10.**3, precision)) + ' KFLOPs'
+        else:
+            return str(flops) + ' FLOPs'
+    else:
+        if units == 'GFLOPs':
+            return str(round(flops / 10.**9, precision)) + ' ' + units
+        elif units == 'MFLOPs':
+            return str(round(flops / 10.**6, precision)) + ' ' + units
+        elif units == 'KFLOPs':
+            return str(round(flops / 10.**3, precision)) + ' ' + units
+        else:
+            return str(flops) + ' FLOPs'
+
+
+def params_to_string(num_params, units=None, precision=2):
+    """Convert parameter number into a string.
+
+    Args:
+        num_params (float): Parameter number to be converted.
+        units (str | None): Converted FLOPs units. Options are None, 'M',
+            'K' and ''. If set to None, it will automatically choose the most
+            suitable unit for Parameter number. Default: None.
+        precision (int): Digit number after the decimal point. Default: 2.
+
+    Returns:
+        str: The converted parameter number with units.
+
+    Examples:
+        >>> params_to_string(1e9)
+        '1000.0 M'
+        >>> params_to_string(2e5)
+        '200.0 k'
+        >>> params_to_string(3e-9)
+        '3e-09'
+    """
+    if units is None:
+        if num_params // 10**6 > 0:
+            return str(round(num_params / 10**6, precision)) + ' M'
+        elif num_params // 10**3:
+            return str(round(num_params / 10**3, precision)) + ' k'
+        else:
+            return str(num_params)
+    else:
+        if units == 'M':
+            return str(round(num_params / 10.**6, precision)) + ' ' + units
+        elif units == 'K':
+            return str(round(num_params / 10.**3, precision)) + ' ' + units
+        else:
+            return str(num_params)
+
+
+def print_model_with_flops(model,
+                           total_flops,
+                           total_params,
+                           units='GFLOPs',
+                           precision=3,
+                           ost=sys.stdout,
+                           flush=False):
+    """Print a model with FLOPs for each layer.
+
+    Args:
+        model (nn.Module): The model to be printed.
+        total_flops (float): Total FLOPs of the model.
+        total_params (float): Total parameter counts of the model.
+        units (str | None): Converted FLOPs units. Default: 'GFLOPs'.
+        precision (int): Digit number after the decimal point. Default: 3.
+        ost (stream): same as `file` param in :func:`print`.
+            Default: sys.stdout.
+        flush (bool): same as that in :func:`print`. Default: False.
+
+    Example:
+        >>> class ExampleModel(nn.Module):
+
+        >>> def __init__(self):
+        >>>     super().__init__()
+        >>>     self.conv1 = nn.Conv2d(3, 8, 3)
+        >>>     self.conv2 = nn.Conv2d(8, 256, 3)
+        >>>     self.conv3 = nn.Conv2d(256, 8, 3)
+        >>>     self.avg_pool = nn.AdaptiveAvgPool2d((1, 1))
+        >>>     self.flatten = nn.Flatten()
+        >>>     self.fc = nn.Linear(8, 1)
+
+        >>> def forward(self, x):
+        >>>     x = self.conv1(x)
+        >>>     x = self.conv2(x)
+        >>>     x = self.conv3(x)
+        >>>     x = self.avg_pool(x)
+        >>>     x = self.flatten(x)
+        >>>     x = self.fc(x)
+        >>>     return x
+
+        >>> model = ExampleModel()
+        >>> x = (3, 16, 16)
+        to print the complexity information state for each layer, you can use
+        >>> get_model_complexity_info(model, x)
+        or directly use
+        >>> print_model_with_flops(model, 4579784.0, 37361)
+        ExampleModel(
+          0.037 M, 100.000% Params, 0.005 GFLOPs, 100.000% FLOPs,
+          (conv1): Conv2d(0.0 M, 0.600% Params, 0.0 GFLOPs, 0.959% FLOPs, 3, 8, kernel_size=(3, 3), stride=(1, 1))  # noqa: E501
+          (conv2): Conv2d(0.019 M, 50.020% Params, 0.003 GFLOPs, 58.760% FLOPs, 8, 256, kernel_size=(3, 3), stride=(1, 1))
+          (conv3): Conv2d(0.018 M, 49.356% Params, 0.002 GFLOPs, 40.264% FLOPs, 256, 8, kernel_size=(3, 3), stride=(1, 1))
+          (avg_pool): AdaptiveAvgPool2d(0.0 M, 0.000% Params, 0.0 GFLOPs, 0.017% FLOPs, output_size=(1, 1))
+          (flatten): Flatten(0.0 M, 0.000% Params, 0.0 GFLOPs, 0.000% FLOPs, )
+          (fc): Linear(0.0 M, 0.024% Params, 0.0 GFLOPs, 0.000% FLOPs, in_features=8, out_features=1, bias=True)
+        )
+    """
+
+    def accumulate_params(self):
+        if is_supported_instance(self):
+            return self.__params__
+        else:
+            sum = 0
+            for m in self.children():
+                sum += m.accumulate_params()
+            return sum
+
+    def accumulate_flops(self):
+        if is_supported_instance(self):
+            return self.__flops__ / model.__batch_counter__
+        else:
+            sum = 0
+            for m in self.children():
+                sum += m.accumulate_flops()
+            return sum
+
+    def flops_repr(self):
+        accumulated_num_params = self.accumulate_params()
+        accumulated_flops_cost = self.accumulate_flops()
+        return ', '.join([
+            params_to_string(
+                accumulated_num_params, units='M', precision=precision),
+            '{:.3%} Params'.format(accumulated_num_params / total_params),
+            flops_to_string(
+                accumulated_flops_cost, units=units, precision=precision),
+            '{:.3%} FLOPs'.format(accumulated_flops_cost / total_flops),
+            self.original_extra_repr()
+        ])
+
+    def add_extra_repr(m):
+        m.accumulate_flops = accumulate_flops.__get__(m)
+        m.accumulate_params = accumulate_params.__get__(m)
+        flops_extra_repr = flops_repr.__get__(m)
+        if m.extra_repr != flops_extra_repr:
+            m.original_extra_repr = m.extra_repr
+            m.extra_repr = flops_extra_repr
+            assert m.extra_repr != m.original_extra_repr
+
+    def del_extra_repr(m):
+        if hasattr(m, 'original_extra_repr'):
+            m.extra_repr = m.original_extra_repr
+            del m.original_extra_repr
+        if hasattr(m, 'accumulate_flops'):
+            del m.accumulate_flops
+
+    model.apply(add_extra_repr)
+    print(model, file=ost, flush=flush)
+    model.apply(del_extra_repr)
+
+
+def get_model_parameters_number(model):
+    """Calculate parameter number of a model.
+
+    Args:
+        model (nn.module): The model for parameter number calculation.
+
+    Returns:
+        float: Parameter number of the model.
+    """
+    num_params = sum(p.numel() for p in model.parameters() if p.requires_grad)
+    return num_params
+
+
+def add_flops_counting_methods(net_main_module):
+    # adding additional methods to the existing module object,
+    # this is done this way so that each function has access to self object
+    net_main_module.start_flops_count = start_flops_count.__get__(
+        net_main_module)
+    net_main_module.stop_flops_count = stop_flops_count.__get__(
+        net_main_module)
+    net_main_module.reset_flops_count = reset_flops_count.__get__(
+        net_main_module)
+    net_main_module.compute_average_flops_cost = compute_average_flops_cost.__get__(  # noqa: E501
+        net_main_module)
+
+    net_main_module.reset_flops_count()
+
+    return net_main_module
+
+
+def compute_average_flops_cost(self):
+    """Compute average FLOPs cost.
+
+    A method to compute average FLOPs cost, which will be available after
+    `add_flops_counting_methods()` is called on a desired net object.
+
+    Returns:
+        float: Current mean flops consumption per image.
+    """
+    batches_count = self.__batch_counter__
+    flops_sum = 0
+    for module in self.modules():
+        if is_supported_instance(module):
+            flops_sum += module.__flops__
+    params_sum = get_model_parameters_number(self)
+    return flops_sum / batches_count, params_sum
+
+
+def start_flops_count(self):
+    """Activate the computation of mean flops consumption per image.
+
+    A method to activate the computation of mean flops consumption per image.
+    which will be available after ``add_flops_counting_methods()`` is called on
+    a desired net object. It should be called before running the network.
+    """
+    add_batch_counter_hook_function(self)
+
+    def add_flops_counter_hook_function(module):
+        if is_supported_instance(module):
+            if hasattr(module, '__flops_handle__'):
+                return
+
+            else:
+                handle = module.register_forward_hook(
+                    get_modules_mapping()[type(module)])
+
+            module.__flops_handle__ = handle
+
+    self.apply(partial(add_flops_counter_hook_function))
+
+
+def stop_flops_count(self):
+    """Stop computing the mean flops consumption per image.
+
+    A method to stop computing the mean flops consumption per image, which will
+    be available after ``add_flops_counting_methods()`` is called on a desired
+    net object. It can be called to pause the computation whenever.
+    """
+    remove_batch_counter_hook_function(self)
+    self.apply(remove_flops_counter_hook_function)
+
+
+def reset_flops_count(self):
+    """Reset statistics computed so far.
+
+    A method to Reset computed statistics, which will be available after
+    `add_flops_counting_methods()` is called on a desired net object.
+    """
+    add_batch_counter_variables_or_reset(self)
+    self.apply(add_flops_counter_variable_or_reset)
+
+
+# ---- Internal functions
+def empty_flops_counter_hook(module, input, output):
+    module.__flops__ += 0
+
+
+def upsample_flops_counter_hook(module, input, output):
+    output_size = output[0]
+    batch_size = output_size.shape[0]
+    output_elements_count = batch_size
+    for val in output_size.shape[1:]:
+        output_elements_count *= val
+    module.__flops__ += int(output_elements_count)
+
+
+def relu_flops_counter_hook(module, input, output):
+    active_elements_count = output.numel()
+    module.__flops__ += int(active_elements_count)
+
+
+def linear_flops_counter_hook(module, input, output):
+    input = input[0]
+    output_last_dim = output.shape[
+        -1]  # pytorch checks dimensions, so here we don't care much
+    module.__flops__ += int(np.prod(input.shape) * output_last_dim)
+
+
+def pool_flops_counter_hook(module, input, output):
+    input = input[0]
+    module.__flops__ += int(np.prod(input.shape))
+
+
+def norm_flops_counter_hook(module, input, output):
+    input = input[0]
+
+    batch_flops = np.prod(input.shape)
+    if (getattr(module, 'affine', False)
+            or getattr(module, 'elementwise_affine', False)):
+        batch_flops *= 2
+    module.__flops__ += int(batch_flops)
+
+
+def deconv_flops_counter_hook(conv_module, input, output):
+    # Can have multiple inputs, getting the first one
+    input = input[0]
+
+    batch_size = input.shape[0]
+    input_height, input_width = input.shape[2:]
+
+    kernel_height, kernel_width = conv_module.kernel_size
+    in_channels = conv_module.in_channels
+    out_channels = conv_module.out_channels
+    groups = conv_module.groups
+
+    filters_per_channel = out_channels // groups
+    conv_per_position_flops = (
+        kernel_height * kernel_width * in_channels * filters_per_channel)
+
+    active_elements_count = batch_size * input_height * input_width
+    overall_conv_flops = conv_per_position_flops * active_elements_count
+    bias_flops = 0
+    if conv_module.bias is not None:
+        output_height, output_width = output.shape[2:]
+        bias_flops = out_channels * batch_size * output_height * output_height
+    overall_flops = overall_conv_flops + bias_flops
+
+    conv_module.__flops__ += int(overall_flops)
+
+
+def conv_flops_counter_hook(conv_module, input, output):
+    # Can have multiple inputs, getting the first one
+    input = input[0]
+
+    batch_size = input.shape[0]
+    output_dims = list(output.shape[2:])
+
+    kernel_dims = list(conv_module.kernel_size)
+    in_channels = conv_module.in_channels
+    out_channels = conv_module.out_channels
+    groups = conv_module.groups
+
+    filters_per_channel = out_channels // groups
+    conv_per_position_flops = int(
+        np.prod(kernel_dims)) * in_channels * filters_per_channel
+
+    active_elements_count = batch_size * int(np.prod(output_dims))
+
+    overall_conv_flops = conv_per_position_flops * active_elements_count
+
+    bias_flops = 0
+
+    if conv_module.bias is not None:
+
+        bias_flops = out_channels * active_elements_count
+
+    overall_flops = overall_conv_flops + bias_flops
+
+    conv_module.__flops__ += int(overall_flops)
+
+
+def batch_counter_hook(module, input, output):
+    batch_size = 1
+    if len(input) > 0:
+        # Can have multiple inputs, getting the first one
+        input = input[0]
+        batch_size = len(input)
+    else:
+        pass
+        print('Warning! No positional inputs found for a module, '
+              'assuming batch size is 1.')
+    module.__batch_counter__ += batch_size
+
+
+def add_batch_counter_variables_or_reset(module):
+
+    module.__batch_counter__ = 0
+
+
+def add_batch_counter_hook_function(module):
+    if hasattr(module, '__batch_counter_handle__'):
+        return
+
+    handle = module.register_forward_hook(batch_counter_hook)
+    module.__batch_counter_handle__ = handle
+
+
+def remove_batch_counter_hook_function(module):
+    if hasattr(module, '__batch_counter_handle__'):
+        module.__batch_counter_handle__.remove()
+        del module.__batch_counter_handle__
+
+
+def add_flops_counter_variable_or_reset(module):
+    if is_supported_instance(module):
+        if hasattr(module, '__flops__') or hasattr(module, '__params__'):
+            print('Warning: variables __flops__ or __params__ are already '
+                  'defined for the module' + type(module).__name__ +
+                  ' ptflops can affect your code!')
+        module.__flops__ = 0
+        module.__params__ = get_model_parameters_number(module)
+
+
+def is_supported_instance(module):
+    if type(module) in get_modules_mapping():
+        return True
+    return False
+
+
+def remove_flops_counter_hook_function(module):
+    if is_supported_instance(module):
+        if hasattr(module, '__flops_handle__'):
+            module.__flops_handle__.remove()
+            del module.__flops_handle__
+
+
+def get_modules_mapping():
+    return {
+        # convolutions
+        nn.Conv1d: conv_flops_counter_hook,
+        nn.Conv2d: conv_flops_counter_hook,
+        mmcv.cnn.bricks.Conv2d: conv_flops_counter_hook,
+        nn.Conv3d: conv_flops_counter_hook,
+        mmcv.cnn.bricks.Conv3d: conv_flops_counter_hook,
+        # activations
+        nn.ReLU: relu_flops_counter_hook,
+        nn.PReLU: relu_flops_counter_hook,
+        nn.ELU: relu_flops_counter_hook,
+        nn.LeakyReLU: relu_flops_counter_hook,
+        nn.ReLU6: relu_flops_counter_hook,
+        # poolings
+        nn.MaxPool1d: pool_flops_counter_hook,
+        nn.AvgPool1d: pool_flops_counter_hook,
+        nn.AvgPool2d: pool_flops_counter_hook,
+        nn.MaxPool2d: pool_flops_counter_hook,
+        mmcv.cnn.bricks.MaxPool2d: pool_flops_counter_hook,
+        nn.MaxPool3d: pool_flops_counter_hook,
+        mmcv.cnn.bricks.MaxPool3d: pool_flops_counter_hook,
+        nn.AvgPool3d: pool_flops_counter_hook,
+        nn.AdaptiveMaxPool1d: pool_flops_counter_hook,
+        nn.AdaptiveAvgPool1d: pool_flops_counter_hook,
+        nn.AdaptiveMaxPool2d: pool_flops_counter_hook,
+        nn.AdaptiveAvgPool2d: pool_flops_counter_hook,
+        nn.AdaptiveMaxPool3d: pool_flops_counter_hook,
+        nn.AdaptiveAvgPool3d: pool_flops_counter_hook,
+        # normalizations
+        nn.BatchNorm1d: norm_flops_counter_hook,
+        nn.BatchNorm2d: norm_flops_counter_hook,
+        nn.BatchNorm3d: norm_flops_counter_hook,
+        nn.GroupNorm: norm_flops_counter_hook,
+        nn.InstanceNorm1d: norm_flops_counter_hook,
+        nn.InstanceNorm2d: norm_flops_counter_hook,
+        nn.InstanceNorm3d: norm_flops_counter_hook,
+        nn.LayerNorm: norm_flops_counter_hook,
+        # FC
+        nn.Linear: linear_flops_counter_hook,
+        mmcv.cnn.bricks.Linear: linear_flops_counter_hook,
+        # Upscale
+        nn.Upsample: upsample_flops_counter_hook,
+        # Deconvolution
+        nn.ConvTranspose2d: deconv_flops_counter_hook,
+        mmcv.cnn.bricks.ConvTranspose2d: deconv_flops_counter_hook,
+    }
diff --git a/annotator/uniformer/mmcv/cnn/utils/fuse_conv_bn.py b/annotator/uniformer/mmcv/cnn/utils/fuse_conv_bn.py
new file mode 100644
index 0000000000000000000000000000000000000000..cb7076f80bf37f7931185bf0293ffcc1ce19c8ef
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/utils/fuse_conv_bn.py
@@ -0,0 +1,59 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+
+
+def _fuse_conv_bn(conv, bn):
+    """Fuse conv and bn into one module.
+
+    Args:
+        conv (nn.Module): Conv to be fused.
+        bn (nn.Module): BN to be fused.
+
+    Returns:
+        nn.Module: Fused module.
+    """
+    conv_w = conv.weight
+    conv_b = conv.bias if conv.bias is not None else torch.zeros_like(
+        bn.running_mean)
+
+    factor = bn.weight / torch.sqrt(bn.running_var + bn.eps)
+    conv.weight = nn.Parameter(conv_w *
+                               factor.reshape([conv.out_channels, 1, 1, 1]))
+    conv.bias = nn.Parameter((conv_b - bn.running_mean) * factor + bn.bias)
+    return conv
+
+
+def fuse_conv_bn(module):
+    """Recursively fuse conv and bn in a module.
+
+    During inference, the functionary of batch norm layers is turned off
+    but only the mean and var alone channels are used, which exposes the
+    chance to fuse it with the preceding conv layers to save computations and
+    simplify network structures.
+
+    Args:
+        module (nn.Module): Module to be fused.
+
+    Returns:
+        nn.Module: Fused module.
+    """
+    last_conv = None
+    last_conv_name = None
+
+    for name, child in module.named_children():
+        if isinstance(child,
+                      (nn.modules.batchnorm._BatchNorm, nn.SyncBatchNorm)):
+            if last_conv is None:  # only fuse BN that is after Conv
+                continue
+            fused_conv = _fuse_conv_bn(last_conv, child)
+            module._modules[last_conv_name] = fused_conv
+            # To reduce changes, set BN as Identity instead of deleting it.
+            module._modules[name] = nn.Identity()
+            last_conv = None
+        elif isinstance(child, nn.Conv2d):
+            last_conv = child
+            last_conv_name = name
+        else:
+            fuse_conv_bn(child)
+    return module
diff --git a/annotator/uniformer/mmcv/cnn/utils/sync_bn.py b/annotator/uniformer/mmcv/cnn/utils/sync_bn.py
new file mode 100644
index 0000000000000000000000000000000000000000..f78f39181d75bb85c53e8c7c8eaf45690e9f0bee
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/utils/sync_bn.py
@@ -0,0 +1,59 @@
+import torch
+
+import annotator.uniformer.mmcv as mmcv
+
+
+class _BatchNormXd(torch.nn.modules.batchnorm._BatchNorm):
+    """A general BatchNorm layer without input dimension check.
+
+    Reproduced from @kapily's work:
+    (https://github.com/pytorch/pytorch/issues/41081#issuecomment-783961547)
+    The only difference between BatchNorm1d, BatchNorm2d, BatchNorm3d, etc
+    is `_check_input_dim` that is designed for tensor sanity checks.
+    The check has been bypassed in this class for the convenience of converting
+    SyncBatchNorm.
+    """
+
+    def _check_input_dim(self, input):
+        return
+
+
+def revert_sync_batchnorm(module):
+    """Helper function to convert all `SyncBatchNorm` (SyncBN) and
+    `mmcv.ops.sync_bn.SyncBatchNorm`(MMSyncBN) layers in the model to
+    `BatchNormXd` layers.
+
+    Adapted from @kapily's work:
+    (https://github.com/pytorch/pytorch/issues/41081#issuecomment-783961547)
+
+    Args:
+        module (nn.Module): The module containing `SyncBatchNorm` layers.
+
+    Returns:
+        module_output: The converted module with `BatchNormXd` layers.
+    """
+    module_output = module
+    module_checklist = [torch.nn.modules.batchnorm.SyncBatchNorm]
+    if hasattr(mmcv, 'ops'):
+        module_checklist.append(mmcv.ops.SyncBatchNorm)
+    if isinstance(module, tuple(module_checklist)):
+        module_output = _BatchNormXd(module.num_features, module.eps,
+                                     module.momentum, module.affine,
+                                     module.track_running_stats)
+        if module.affine:
+            # no_grad() may not be needed here but
+            # just to be consistent with `convert_sync_batchnorm()`
+            with torch.no_grad():
+                module_output.weight = module.weight
+                module_output.bias = module.bias
+        module_output.running_mean = module.running_mean
+        module_output.running_var = module.running_var
+        module_output.num_batches_tracked = module.num_batches_tracked
+        module_output.training = module.training
+        # qconfig exists in quantized models
+        if hasattr(module, 'qconfig'):
+            module_output.qconfig = module.qconfig
+    for name, child in module.named_children():
+        module_output.add_module(name, revert_sync_batchnorm(child))
+    del module
+    return module_output
diff --git a/annotator/uniformer/mmcv/cnn/utils/weight_init.py b/annotator/uniformer/mmcv/cnn/utils/weight_init.py
new file mode 100644
index 0000000000000000000000000000000000000000..287a1d0bffe26e023029d48634d9b761deda7ba4
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/utils/weight_init.py
@@ -0,0 +1,684 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import copy
+import math
+import warnings
+
+import numpy as np
+import torch
+import torch.nn as nn
+from torch import Tensor
+
+from annotator.uniformer.mmcv.utils import Registry, build_from_cfg, get_logger, print_log
+
+INITIALIZERS = Registry('initializer')
+
+
+def update_init_info(module, init_info):
+    """Update the `_params_init_info` in the module if the value of parameters
+    are changed.
+
+    Args:
+        module (obj:`nn.Module`): The module of PyTorch with a user-defined
+            attribute `_params_init_info` which records the initialization
+            information.
+        init_info (str): The string that describes the initialization.
+    """
+    assert hasattr(
+        module,
+        '_params_init_info'), f'Can not find `_params_init_info` in {module}'
+    for name, param in module.named_parameters():
+
+        assert param in module._params_init_info, (
+            f'Find a new :obj:`Parameter` '
+            f'named `{name}` during executing the '
+            f'`init_weights` of '
+            f'`{module.__class__.__name__}`. '
+            f'Please do not add or '
+            f'replace parameters during executing '
+            f'the `init_weights`. ')
+
+        # The parameter has been changed during executing the
+        # `init_weights` of module
+        mean_value = param.data.mean()
+        if module._params_init_info[param]['tmp_mean_value'] != mean_value:
+            module._params_init_info[param]['init_info'] = init_info
+            module._params_init_info[param]['tmp_mean_value'] = mean_value
+
+
+def constant_init(module, val, bias=0):
+    if hasattr(module, 'weight') and module.weight is not None:
+        nn.init.constant_(module.weight, val)
+    if hasattr(module, 'bias') and module.bias is not None:
+        nn.init.constant_(module.bias, bias)
+
+
+def xavier_init(module, gain=1, bias=0, distribution='normal'):
+    assert distribution in ['uniform', 'normal']
+    if hasattr(module, 'weight') and module.weight is not None:
+        if distribution == 'uniform':
+            nn.init.xavier_uniform_(module.weight, gain=gain)
+        else:
+            nn.init.xavier_normal_(module.weight, gain=gain)
+    if hasattr(module, 'bias') and module.bias is not None:
+        nn.init.constant_(module.bias, bias)
+
+
+def normal_init(module, mean=0, std=1, bias=0):
+    if hasattr(module, 'weight') and module.weight is not None:
+        nn.init.normal_(module.weight, mean, std)
+    if hasattr(module, 'bias') and module.bias is not None:
+        nn.init.constant_(module.bias, bias)
+
+
+def trunc_normal_init(module: nn.Module,
+                      mean: float = 0,
+                      std: float = 1,
+                      a: float = -2,
+                      b: float = 2,
+                      bias: float = 0) -> None:
+    if hasattr(module, 'weight') and module.weight is not None:
+        trunc_normal_(module.weight, mean, std, a, b)  # type: ignore
+    if hasattr(module, 'bias') and module.bias is not None:
+        nn.init.constant_(module.bias, bias)  # type: ignore
+
+
+def uniform_init(module, a=0, b=1, bias=0):
+    if hasattr(module, 'weight') and module.weight is not None:
+        nn.init.uniform_(module.weight, a, b)
+    if hasattr(module, 'bias') and module.bias is not None:
+        nn.init.constant_(module.bias, bias)
+
+
+def kaiming_init(module,
+                 a=0,
+                 mode='fan_out',
+                 nonlinearity='relu',
+                 bias=0,
+                 distribution='normal'):
+    assert distribution in ['uniform', 'normal']
+    if hasattr(module, 'weight') and module.weight is not None:
+        if distribution == 'uniform':
+            nn.init.kaiming_uniform_(
+                module.weight, a=a, mode=mode, nonlinearity=nonlinearity)
+        else:
+            nn.init.kaiming_normal_(
+                module.weight, a=a, mode=mode, nonlinearity=nonlinearity)
+    if hasattr(module, 'bias') and module.bias is not None:
+        nn.init.constant_(module.bias, bias)
+
+
+def caffe2_xavier_init(module, bias=0):
+    # `XavierFill` in Caffe2 corresponds to `kaiming_uniform_` in PyTorch
+    # Acknowledgment to FAIR's internal code
+    kaiming_init(
+        module,
+        a=1,
+        mode='fan_in',
+        nonlinearity='leaky_relu',
+        bias=bias,
+        distribution='uniform')
+
+
+def bias_init_with_prob(prior_prob):
+    """initialize conv/fc bias value according to a given probability value."""
+    bias_init = float(-np.log((1 - prior_prob) / prior_prob))
+    return bias_init
+
+
+def _get_bases_name(m):
+    return [b.__name__ for b in m.__class__.__bases__]
+
+
+class BaseInit(object):
+
+    def __init__(self, *, bias=0, bias_prob=None, layer=None):
+        self.wholemodule = False
+        if not isinstance(bias, (int, float)):
+            raise TypeError(f'bias must be a number, but got a {type(bias)}')
+
+        if bias_prob is not None:
+            if not isinstance(bias_prob, float):
+                raise TypeError(f'bias_prob type must be float, \
+                    but got {type(bias_prob)}')
+
+        if layer is not None:
+            if not isinstance(layer, (str, list)):
+                raise TypeError(f'layer must be a str or a list of str, \
+                    but got a {type(layer)}')
+        else:
+            layer = []
+
+        if bias_prob is not None:
+            self.bias = bias_init_with_prob(bias_prob)
+        else:
+            self.bias = bias
+        self.layer = [layer] if isinstance(layer, str) else layer
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}, bias={self.bias}'
+        return info
+
+
+@INITIALIZERS.register_module(name='Constant')
+class ConstantInit(BaseInit):
+    """Initialize module parameters with constant values.
+
+    Args:
+        val (int | float): the value to fill the weights in the module with
+        bias (int | float): the value to fill the bias. Defaults to 0.
+        bias_prob (float, optional): the probability for bias initialization.
+            Defaults to None.
+        layer (str | list[str], optional): the layer will be initialized.
+            Defaults to None.
+    """
+
+    def __init__(self, val, **kwargs):
+        super().__init__(**kwargs)
+        self.val = val
+
+    def __call__(self, module):
+
+        def init(m):
+            if self.wholemodule:
+                constant_init(m, self.val, self.bias)
+            else:
+                layername = m.__class__.__name__
+                basesname = _get_bases_name(m)
+                if len(set(self.layer) & set([layername] + basesname)):
+                    constant_init(m, self.val, self.bias)
+
+        module.apply(init)
+        if hasattr(module, '_params_init_info'):
+            update_init_info(module, init_info=self._get_init_info())
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}: val={self.val}, bias={self.bias}'
+        return info
+
+
+@INITIALIZERS.register_module(name='Xavier')
+class XavierInit(BaseInit):
+    r"""Initialize module parameters with values according to the method
+    described in `Understanding the difficulty of training deep feedforward
+    neural networks - Glorot, X. & Bengio, Y. (2010).
+    <http://proceedings.mlr.press/v9/glorot10a/glorot10a.pdf>`_
+
+    Args:
+        gain (int | float): an optional scaling factor. Defaults to 1.
+        bias (int | float): the value to fill the bias. Defaults to 0.
+        bias_prob (float, optional): the probability for bias initialization.
+            Defaults to None.
+        distribution (str): distribution either be ``'normal'``
+            or ``'uniform'``. Defaults to ``'normal'``.
+        layer (str | list[str], optional): the layer will be initialized.
+            Defaults to None.
+    """
+
+    def __init__(self, gain=1, distribution='normal', **kwargs):
+        super().__init__(**kwargs)
+        self.gain = gain
+        self.distribution = distribution
+
+    def __call__(self, module):
+
+        def init(m):
+            if self.wholemodule:
+                xavier_init(m, self.gain, self.bias, self.distribution)
+            else:
+                layername = m.__class__.__name__
+                basesname = _get_bases_name(m)
+                if len(set(self.layer) & set([layername] + basesname)):
+                    xavier_init(m, self.gain, self.bias, self.distribution)
+
+        module.apply(init)
+        if hasattr(module, '_params_init_info'):
+            update_init_info(module, init_info=self._get_init_info())
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}: gain={self.gain}, ' \
+               f'distribution={self.distribution}, bias={self.bias}'
+        return info
+
+
+@INITIALIZERS.register_module(name='Normal')
+class NormalInit(BaseInit):
+    r"""Initialize module parameters with the values drawn from the normal
+    distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)`.
+
+    Args:
+        mean (int | float):the mean of the normal distribution. Defaults to 0.
+        std (int | float): the standard deviation of the normal distribution.
+            Defaults to 1.
+        bias (int | float): the value to fill the bias. Defaults to 0.
+        bias_prob (float, optional): the probability for bias initialization.
+            Defaults to None.
+        layer (str | list[str], optional): the layer will be initialized.
+            Defaults to None.
+
+    """
+
+    def __init__(self, mean=0, std=1, **kwargs):
+        super().__init__(**kwargs)
+        self.mean = mean
+        self.std = std
+
+    def __call__(self, module):
+
+        def init(m):
+            if self.wholemodule:
+                normal_init(m, self.mean, self.std, self.bias)
+            else:
+                layername = m.__class__.__name__
+                basesname = _get_bases_name(m)
+                if len(set(self.layer) & set([layername] + basesname)):
+                    normal_init(m, self.mean, self.std, self.bias)
+
+        module.apply(init)
+        if hasattr(module, '_params_init_info'):
+            update_init_info(module, init_info=self._get_init_info())
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}: mean={self.mean},' \
+               f' std={self.std}, bias={self.bias}'
+        return info
+
+
+@INITIALIZERS.register_module(name='TruncNormal')
+class TruncNormalInit(BaseInit):
+    r"""Initialize module parameters with the values drawn from the normal
+    distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)` with values
+    outside :math:`[a, b]`.
+
+    Args:
+        mean (float): the mean of the normal distribution. Defaults to 0.
+        std (float):  the standard deviation of the normal distribution.
+            Defaults to 1.
+        a (float): The minimum cutoff value.
+        b ( float): The maximum cutoff value.
+        bias (float): the value to fill the bias. Defaults to 0.
+        bias_prob (float, optional): the probability for bias initialization.
+            Defaults to None.
+        layer (str | list[str], optional): the layer will be initialized.
+            Defaults to None.
+
+    """
+
+    def __init__(self,
+                 mean: float = 0,
+                 std: float = 1,
+                 a: float = -2,
+                 b: float = 2,
+                 **kwargs) -> None:
+        super().__init__(**kwargs)
+        self.mean = mean
+        self.std = std
+        self.a = a
+        self.b = b
+
+    def __call__(self, module: nn.Module) -> None:
+
+        def init(m):
+            if self.wholemodule:
+                trunc_normal_init(m, self.mean, self.std, self.a, self.b,
+                                  self.bias)
+            else:
+                layername = m.__class__.__name__
+                basesname = _get_bases_name(m)
+                if len(set(self.layer) & set([layername] + basesname)):
+                    trunc_normal_init(m, self.mean, self.std, self.a, self.b,
+                                      self.bias)
+
+        module.apply(init)
+        if hasattr(module, '_params_init_info'):
+            update_init_info(module, init_info=self._get_init_info())
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}: a={self.a}, b={self.b},' \
+               f' mean={self.mean}, std={self.std}, bias={self.bias}'
+        return info
+
+
+@INITIALIZERS.register_module(name='Uniform')
+class UniformInit(BaseInit):
+    r"""Initialize module parameters with values drawn from the uniform
+    distribution :math:`\mathcal{U}(a, b)`.
+
+    Args:
+        a (int | float): the lower bound of the uniform distribution.
+            Defaults to 0.
+        b (int | float): the upper bound of the uniform distribution.
+            Defaults to 1.
+        bias (int | float): the value to fill the bias. Defaults to 0.
+        bias_prob (float, optional): the probability for bias initialization.
+            Defaults to None.
+        layer (str | list[str], optional): the layer will be initialized.
+            Defaults to None.
+    """
+
+    def __init__(self, a=0, b=1, **kwargs):
+        super().__init__(**kwargs)
+        self.a = a
+        self.b = b
+
+    def __call__(self, module):
+
+        def init(m):
+            if self.wholemodule:
+                uniform_init(m, self.a, self.b, self.bias)
+            else:
+                layername = m.__class__.__name__
+                basesname = _get_bases_name(m)
+                if len(set(self.layer) & set([layername] + basesname)):
+                    uniform_init(m, self.a, self.b, self.bias)
+
+        module.apply(init)
+        if hasattr(module, '_params_init_info'):
+            update_init_info(module, init_info=self._get_init_info())
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}: a={self.a},' \
+               f' b={self.b}, bias={self.bias}'
+        return info
+
+
+@INITIALIZERS.register_module(name='Kaiming')
+class KaimingInit(BaseInit):
+    r"""Initialize module parameters with the values according to the method
+    described in `Delving deep into rectifiers: Surpassing human-level
+    performance on ImageNet classification - He, K. et al. (2015).
+    <https://www.cv-foundation.org/openaccess/content_iccv_2015/
+    papers/He_Delving_Deep_into_ICCV_2015_paper.pdf>`_
+
+    Args:
+        a (int | float): the negative slope of the rectifier used after this
+            layer (only used with ``'leaky_relu'``). Defaults to 0.
+        mode (str):  either ``'fan_in'`` or ``'fan_out'``. Choosing
+            ``'fan_in'`` preserves the magnitude of the variance of the weights
+            in the forward pass. Choosing ``'fan_out'`` preserves the
+            magnitudes in the backwards pass. Defaults to ``'fan_out'``.
+        nonlinearity (str): the non-linear function (`nn.functional` name),
+            recommended to use only with ``'relu'`` or ``'leaky_relu'`` .
+            Defaults to 'relu'.
+        bias (int | float): the value to fill the bias. Defaults to 0.
+        bias_prob (float, optional): the probability for bias initialization.
+            Defaults to None.
+        distribution (str): distribution either be ``'normal'`` or
+            ``'uniform'``. Defaults to ``'normal'``.
+        layer (str | list[str], optional): the layer will be initialized.
+            Defaults to None.
+    """
+
+    def __init__(self,
+                 a=0,
+                 mode='fan_out',
+                 nonlinearity='relu',
+                 distribution='normal',
+                 **kwargs):
+        super().__init__(**kwargs)
+        self.a = a
+        self.mode = mode
+        self.nonlinearity = nonlinearity
+        self.distribution = distribution
+
+    def __call__(self, module):
+
+        def init(m):
+            if self.wholemodule:
+                kaiming_init(m, self.a, self.mode, self.nonlinearity,
+                             self.bias, self.distribution)
+            else:
+                layername = m.__class__.__name__
+                basesname = _get_bases_name(m)
+                if len(set(self.layer) & set([layername] + basesname)):
+                    kaiming_init(m, self.a, self.mode, self.nonlinearity,
+                                 self.bias, self.distribution)
+
+        module.apply(init)
+        if hasattr(module, '_params_init_info'):
+            update_init_info(module, init_info=self._get_init_info())
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}: a={self.a}, mode={self.mode}, ' \
+               f'nonlinearity={self.nonlinearity}, ' \
+               f'distribution ={self.distribution}, bias={self.bias}'
+        return info
+
+
+@INITIALIZERS.register_module(name='Caffe2Xavier')
+class Caffe2XavierInit(KaimingInit):
+    # `XavierFill` in Caffe2 corresponds to `kaiming_uniform_` in PyTorch
+    # Acknowledgment to FAIR's internal code
+    def __init__(self, **kwargs):
+        super().__init__(
+            a=1,
+            mode='fan_in',
+            nonlinearity='leaky_relu',
+            distribution='uniform',
+            **kwargs)
+
+    def __call__(self, module):
+        super().__call__(module)
+
+
+@INITIALIZERS.register_module(name='Pretrained')
+class PretrainedInit(object):
+    """Initialize module by loading a pretrained model.
+
+    Args:
+        checkpoint (str): the checkpoint file of the pretrained model should
+            be load.
+        prefix (str, optional): the prefix of a sub-module in the pretrained
+            model. it is for loading a part of the pretrained model to
+            initialize. For example, if we would like to only load the
+            backbone of a detector model, we can set ``prefix='backbone.'``.
+            Defaults to None.
+        map_location (str): map tensors into proper locations.
+    """
+
+    def __init__(self, checkpoint, prefix=None, map_location=None):
+        self.checkpoint = checkpoint
+        self.prefix = prefix
+        self.map_location = map_location
+
+    def __call__(self, module):
+        from annotator.uniformer.mmcv.runner import (_load_checkpoint_with_prefix, load_checkpoint,
+                                 load_state_dict)
+        logger = get_logger('mmcv')
+        if self.prefix is None:
+            print_log(f'load model from: {self.checkpoint}', logger=logger)
+            load_checkpoint(
+                module,
+                self.checkpoint,
+                map_location=self.map_location,
+                strict=False,
+                logger=logger)
+        else:
+            print_log(
+                f'load {self.prefix} in model from: {self.checkpoint}',
+                logger=logger)
+            state_dict = _load_checkpoint_with_prefix(
+                self.prefix, self.checkpoint, map_location=self.map_location)
+            load_state_dict(module, state_dict, strict=False, logger=logger)
+
+        if hasattr(module, '_params_init_info'):
+            update_init_info(module, init_info=self._get_init_info())
+
+    def _get_init_info(self):
+        info = f'{self.__class__.__name__}: load from {self.checkpoint}'
+        return info
+
+
+def _initialize(module, cfg, wholemodule=False):
+    func = build_from_cfg(cfg, INITIALIZERS)
+    # wholemodule flag is for override mode, there is no layer key in override
+    # and initializer will give init values for the whole module with the name
+    # in override.
+    func.wholemodule = wholemodule
+    func(module)
+
+
+def _initialize_override(module, override, cfg):
+    if not isinstance(override, (dict, list)):
+        raise TypeError(f'override must be a dict or a list of dict, \
+                but got {type(override)}')
+
+    override = [override] if isinstance(override, dict) else override
+
+    for override_ in override:
+
+        cp_override = copy.deepcopy(override_)
+        name = cp_override.pop('name', None)
+        if name is None:
+            raise ValueError('`override` must contain the key "name",'
+                             f'but got {cp_override}')
+        # if override only has name key, it means use args in init_cfg
+        if not cp_override:
+            cp_override.update(cfg)
+        # if override has name key and other args except type key, it will
+        # raise error
+        elif 'type' not in cp_override.keys():
+            raise ValueError(
+                f'`override` need "type" key, but got {cp_override}')
+
+        if hasattr(module, name):
+            _initialize(getattr(module, name), cp_override, wholemodule=True)
+        else:
+            raise RuntimeError(f'module did not have attribute {name}, '
+                               f'but init_cfg is {cp_override}.')
+
+
+def initialize(module, init_cfg):
+    """Initialize a module.
+
+    Args:
+        module (``torch.nn.Module``): the module will be initialized.
+        init_cfg (dict | list[dict]): initialization configuration dict to
+            define initializer. OpenMMLab has implemented 6 initializers
+            including ``Constant``, ``Xavier``, ``Normal``, ``Uniform``,
+            ``Kaiming``, and ``Pretrained``.
+    Example:
+        >>> module = nn.Linear(2, 3, bias=True)
+        >>> init_cfg = dict(type='Constant', layer='Linear', val =1 , bias =2)
+        >>> initialize(module, init_cfg)
+
+        >>> module = nn.Sequential(nn.Conv1d(3, 1, 3), nn.Linear(1,2))
+        >>> # define key ``'layer'`` for initializing layer with different
+        >>> # configuration
+        >>> init_cfg = [dict(type='Constant', layer='Conv1d', val=1),
+                dict(type='Constant', layer='Linear', val=2)]
+        >>> initialize(module, init_cfg)
+
+        >>> # define key``'override'`` to initialize some specific part in
+        >>> # module
+        >>> class FooNet(nn.Module):
+        >>>     def __init__(self):
+        >>>         super().__init__()
+        >>>         self.feat = nn.Conv2d(3, 16, 3)
+        >>>         self.reg = nn.Conv2d(16, 10, 3)
+        >>>         self.cls = nn.Conv2d(16, 5, 3)
+        >>> model = FooNet()
+        >>> init_cfg = dict(type='Constant', val=1, bias=2, layer='Conv2d',
+        >>>     override=dict(type='Constant', name='reg', val=3, bias=4))
+        >>> initialize(model, init_cfg)
+
+        >>> model = ResNet(depth=50)
+        >>> # Initialize weights with the pretrained model.
+        >>> init_cfg = dict(type='Pretrained',
+                checkpoint='torchvision://resnet50')
+        >>> initialize(model, init_cfg)
+
+        >>> # Initialize weights of a sub-module with the specific part of
+        >>> # a pretrained model by using "prefix".
+        >>> url = 'http://download.openmmlab.com/mmdetection/v2.0/retinanet/'\
+        >>>     'retinanet_r50_fpn_1x_coco/'\
+        >>>     'retinanet_r50_fpn_1x_coco_20200130-c2398f9e.pth'
+        >>> init_cfg = dict(type='Pretrained',
+                checkpoint=url, prefix='backbone.')
+    """
+    if not isinstance(init_cfg, (dict, list)):
+        raise TypeError(f'init_cfg must be a dict or a list of dict, \
+                but got {type(init_cfg)}')
+
+    if isinstance(init_cfg, dict):
+        init_cfg = [init_cfg]
+
+    for cfg in init_cfg:
+        # should deeply copy the original config because cfg may be used by
+        # other modules, e.g., one init_cfg shared by multiple bottleneck
+        # blocks, the expected cfg will be changed after pop and will change
+        # the initialization behavior of other modules
+        cp_cfg = copy.deepcopy(cfg)
+        override = cp_cfg.pop('override', None)
+        _initialize(module, cp_cfg)
+
+        if override is not None:
+            cp_cfg.pop('layer', None)
+            _initialize_override(module, override, cp_cfg)
+        else:
+            # All attributes in module have same initialization.
+            pass
+
+
+def _no_grad_trunc_normal_(tensor: Tensor, mean: float, std: float, a: float,
+                           b: float) -> Tensor:
+    # Method based on
+    # https://people.sc.fsu.edu/~jburkardt/presentations/truncated_normal.pdf
+    # Modified from
+    # https://github.com/pytorch/pytorch/blob/master/torch/nn/init.py
+    def norm_cdf(x):
+        # Computes standard normal cumulative distribution function
+        return (1. + math.erf(x / math.sqrt(2.))) / 2.
+
+    if (mean < a - 2 * std) or (mean > b + 2 * std):
+        warnings.warn(
+            'mean is more than 2 std from [a, b] in nn.init.trunc_normal_. '
+            'The distribution of values may be incorrect.',
+            stacklevel=2)
+
+    with torch.no_grad():
+        # Values are generated by using a truncated uniform distribution and
+        # then using the inverse CDF for the normal distribution.
+        # Get upper and lower cdf values
+        lower = norm_cdf((a - mean) / std)
+        upper = norm_cdf((b - mean) / std)
+
+        # Uniformly fill tensor with values from [lower, upper], then translate
+        # to [2lower-1, 2upper-1].
+        tensor.uniform_(2 * lower - 1, 2 * upper - 1)
+
+        # Use inverse cdf transform for normal distribution to get truncated
+        # standard normal
+        tensor.erfinv_()
+
+        # Transform to proper mean, std
+        tensor.mul_(std * math.sqrt(2.))
+        tensor.add_(mean)
+
+        # Clamp to ensure it's in the proper range
+        tensor.clamp_(min=a, max=b)
+        return tensor
+
+
+def trunc_normal_(tensor: Tensor,
+                  mean: float = 0.,
+                  std: float = 1.,
+                  a: float = -2.,
+                  b: float = 2.) -> Tensor:
+    r"""Fills the input Tensor with values drawn from a truncated
+    normal distribution. The values are effectively drawn from the
+    normal distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)`
+    with values outside :math:`[a, b]` redrawn until they are within
+    the bounds. The method used for generating the random values works
+    best when :math:`a \leq \text{mean} \leq b`.
+
+    Modified from
+    https://github.com/pytorch/pytorch/blob/master/torch/nn/init.py
+
+    Args:
+        tensor (``torch.Tensor``): an n-dimensional `torch.Tensor`.
+        mean (float): the mean of the normal distribution.
+        std (float): the standard deviation of the normal distribution.
+        a (float): the minimum cutoff value.
+        b (float): the maximum cutoff value.
+    """
+    return _no_grad_trunc_normal_(tensor, mean, std, a, b)
diff --git a/annotator/uniformer/mmcv/cnn/vgg.py b/annotator/uniformer/mmcv/cnn/vgg.py
new file mode 100644
index 0000000000000000000000000000000000000000..8778b649561a45a9652b1a15a26c2d171e58f3e1
--- /dev/null
+++ b/annotator/uniformer/mmcv/cnn/vgg.py
@@ -0,0 +1,175 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import logging
+
+import torch.nn as nn
+
+from .utils import constant_init, kaiming_init, normal_init
+
+
+def conv3x3(in_planes, out_planes, dilation=1):
+    """3x3 convolution with padding."""
+    return nn.Conv2d(
+        in_planes,
+        out_planes,
+        kernel_size=3,
+        padding=dilation,
+        dilation=dilation)
+
+
+def make_vgg_layer(inplanes,
+                   planes,
+                   num_blocks,
+                   dilation=1,
+                   with_bn=False,
+                   ceil_mode=False):
+    layers = []
+    for _ in range(num_blocks):
+        layers.append(conv3x3(inplanes, planes, dilation))
+        if with_bn:
+            layers.append(nn.BatchNorm2d(planes))
+        layers.append(nn.ReLU(inplace=True))
+        inplanes = planes
+    layers.append(nn.MaxPool2d(kernel_size=2, stride=2, ceil_mode=ceil_mode))
+
+    return layers
+
+
+class VGG(nn.Module):
+    """VGG backbone.
+
+    Args:
+        depth (int): Depth of vgg, from {11, 13, 16, 19}.
+        with_bn (bool): Use BatchNorm or not.
+        num_classes (int): number of classes for classification.
+        num_stages (int): VGG stages, normally 5.
+        dilations (Sequence[int]): Dilation of each stage.
+        out_indices (Sequence[int]): Output from which stages.
+        frozen_stages (int): Stages to be frozen (all param fixed). -1 means
+            not freezing any parameters.
+        bn_eval (bool): Whether to set BN layers as eval mode, namely, freeze
+            running stats (mean and var).
+        bn_frozen (bool): Whether to freeze weight and bias of BN layers.
+    """
+
+    arch_settings = {
+        11: (1, 1, 2, 2, 2),
+        13: (2, 2, 2, 2, 2),
+        16: (2, 2, 3, 3, 3),
+        19: (2, 2, 4, 4, 4)
+    }
+
+    def __init__(self,
+                 depth,
+                 with_bn=False,
+                 num_classes=-1,
+                 num_stages=5,
+                 dilations=(1, 1, 1, 1, 1),
+                 out_indices=(0, 1, 2, 3, 4),
+                 frozen_stages=-1,
+                 bn_eval=True,
+                 bn_frozen=False,
+                 ceil_mode=False,
+                 with_last_pool=True):
+        super(VGG, self).__init__()
+        if depth not in self.arch_settings:
+            raise KeyError(f'invalid depth {depth} for vgg')
+        assert num_stages >= 1 and num_stages <= 5
+        stage_blocks = self.arch_settings[depth]
+        self.stage_blocks = stage_blocks[:num_stages]
+        assert len(dilations) == num_stages
+        assert max(out_indices) <= num_stages
+
+        self.num_classes = num_classes
+        self.out_indices = out_indices
+        self.frozen_stages = frozen_stages
+        self.bn_eval = bn_eval
+        self.bn_frozen = bn_frozen
+
+        self.inplanes = 3
+        start_idx = 0
+        vgg_layers = []
+        self.range_sub_modules = []
+        for i, num_blocks in enumerate(self.stage_blocks):
+            num_modules = num_blocks * (2 + with_bn) + 1
+            end_idx = start_idx + num_modules
+            dilation = dilations[i]
+            planes = 64 * 2**i if i < 4 else 512
+            vgg_layer = make_vgg_layer(
+                self.inplanes,
+                planes,
+                num_blocks,
+                dilation=dilation,
+                with_bn=with_bn,
+                ceil_mode=ceil_mode)
+            vgg_layers.extend(vgg_layer)
+            self.inplanes = planes
+            self.range_sub_modules.append([start_idx, end_idx])
+            start_idx = end_idx
+        if not with_last_pool:
+            vgg_layers.pop(-1)
+            self.range_sub_modules[-1][1] -= 1
+        self.module_name = 'features'
+        self.add_module(self.module_name, nn.Sequential(*vgg_layers))
+
+        if self.num_classes > 0:
+            self.classifier = nn.Sequential(
+                nn.Linear(512 * 7 * 7, 4096),
+                nn.ReLU(True),
+                nn.Dropout(),
+                nn.Linear(4096, 4096),
+                nn.ReLU(True),
+                nn.Dropout(),
+                nn.Linear(4096, num_classes),
+            )
+
+    def init_weights(self, pretrained=None):
+        if isinstance(pretrained, str):
+            logger = logging.getLogger()
+            from ..runner import load_checkpoint
+            load_checkpoint(self, pretrained, strict=False, logger=logger)
+        elif pretrained is None:
+            for m in self.modules():
+                if isinstance(m, nn.Conv2d):
+                    kaiming_init(m)
+                elif isinstance(m, nn.BatchNorm2d):
+                    constant_init(m, 1)
+                elif isinstance(m, nn.Linear):
+                    normal_init(m, std=0.01)
+        else:
+            raise TypeError('pretrained must be a str or None')
+
+    def forward(self, x):
+        outs = []
+        vgg_layers = getattr(self, self.module_name)
+        for i in range(len(self.stage_blocks)):
+            for j in range(*self.range_sub_modules[i]):
+                vgg_layer = vgg_layers[j]
+                x = vgg_layer(x)
+            if i in self.out_indices:
+                outs.append(x)
+        if self.num_classes > 0:
+            x = x.view(x.size(0), -1)
+            x = self.classifier(x)
+            outs.append(x)
+        if len(outs) == 1:
+            return outs[0]
+        else:
+            return tuple(outs)
+
+    def train(self, mode=True):
+        super(VGG, self).train(mode)
+        if self.bn_eval:
+            for m in self.modules():
+                if isinstance(m, nn.BatchNorm2d):
+                    m.eval()
+                    if self.bn_frozen:
+                        for params in m.parameters():
+                            params.requires_grad = False
+        vgg_layers = getattr(self, self.module_name)
+        if mode and self.frozen_stages >= 0:
+            for i in range(self.frozen_stages):
+                for j in range(*self.range_sub_modules[i]):
+                    mod = vgg_layers[j]
+                    mod.eval()
+                    for param in mod.parameters():
+                        param.requires_grad = False
diff --git a/annotator/uniformer/mmcv/engine/__init__.py b/annotator/uniformer/mmcv/engine/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..3193b7f664e19ce2458d81c836597fa22e4bb082
--- /dev/null
+++ b/annotator/uniformer/mmcv/engine/__init__.py
@@ -0,0 +1,8 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .test import (collect_results_cpu, collect_results_gpu, multi_gpu_test,
+                   single_gpu_test)
+
+__all__ = [
+    'collect_results_cpu', 'collect_results_gpu', 'multi_gpu_test',
+    'single_gpu_test'
+]
diff --git a/annotator/uniformer/mmcv/engine/test.py b/annotator/uniformer/mmcv/engine/test.py
new file mode 100644
index 0000000000000000000000000000000000000000..8dbeef271db634ec2dadfda3bc0b5ef9c7a677ff
--- /dev/null
+++ b/annotator/uniformer/mmcv/engine/test.py
@@ -0,0 +1,202 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import os.path as osp
+import pickle
+import shutil
+import tempfile
+import time
+
+import torch
+import torch.distributed as dist
+
+import annotator.uniformer.mmcv as mmcv
+from annotator.uniformer.mmcv.runner import get_dist_info
+
+
+def single_gpu_test(model, data_loader):
+    """Test model with a single gpu.
+
+    This method tests model with a single gpu and displays test progress bar.
+
+    Args:
+        model (nn.Module): Model to be tested.
+        data_loader (nn.Dataloader): Pytorch data loader.
+
+    Returns:
+        list: The prediction results.
+    """
+    model.eval()
+    results = []
+    dataset = data_loader.dataset
+    prog_bar = mmcv.ProgressBar(len(dataset))
+    for data in data_loader:
+        with torch.no_grad():
+            result = model(return_loss=False, **data)
+        results.extend(result)
+
+        # Assume result has the same length of batch_size
+        # refer to https://github.com/open-mmlab/mmcv/issues/985
+        batch_size = len(result)
+        for _ in range(batch_size):
+            prog_bar.update()
+    return results
+
+
+def multi_gpu_test(model, data_loader, tmpdir=None, gpu_collect=False):
+    """Test model with multiple gpus.
+
+    This method tests model with multiple gpus and collects the results
+    under two different modes: gpu and cpu modes. By setting
+    ``gpu_collect=True``, it encodes results to gpu tensors and use gpu
+    communication for results collection. On cpu mode it saves the results on
+    different gpus to ``tmpdir`` and collects them by the rank 0 worker.
+
+    Args:
+        model (nn.Module): Model to be tested.
+        data_loader (nn.Dataloader): Pytorch data loader.
+        tmpdir (str): Path of directory to save the temporary results from
+            different gpus under cpu mode.
+        gpu_collect (bool): Option to use either gpu or cpu to collect results.
+
+    Returns:
+        list: The prediction results.
+    """
+    model.eval()
+    results = []
+    dataset = data_loader.dataset
+    rank, world_size = get_dist_info()
+    if rank == 0:
+        prog_bar = mmcv.ProgressBar(len(dataset))
+    time.sleep(2)  # This line can prevent deadlock problem in some cases.
+    for i, data in enumerate(data_loader):
+        with torch.no_grad():
+            result = model(return_loss=False, **data)
+        results.extend(result)
+
+        if rank == 0:
+            batch_size = len(result)
+            batch_size_all = batch_size * world_size
+            if batch_size_all + prog_bar.completed > len(dataset):
+                batch_size_all = len(dataset) - prog_bar.completed
+            for _ in range(batch_size_all):
+                prog_bar.update()
+
+    # collect results from all ranks
+    if gpu_collect:
+        results = collect_results_gpu(results, len(dataset))
+    else:
+        results = collect_results_cpu(results, len(dataset), tmpdir)
+    return results
+
+
+def collect_results_cpu(result_part, size, tmpdir=None):
+    """Collect results under cpu mode.
+
+    On cpu mode, this function will save the results on different gpus to
+    ``tmpdir`` and collect them by the rank 0 worker.
+
+    Args:
+        result_part (list): Result list containing result parts
+            to be collected.
+        size (int): Size of the results, commonly equal to length of
+            the results.
+        tmpdir (str | None): temporal directory for collected results to
+            store. If set to None, it will create a random temporal directory
+            for it.
+
+    Returns:
+        list: The collected results.
+    """
+    rank, world_size = get_dist_info()
+    # create a tmp dir if it is not specified
+    if tmpdir is None:
+        MAX_LEN = 512
+        # 32 is whitespace
+        dir_tensor = torch.full((MAX_LEN, ),
+                                32,
+                                dtype=torch.uint8,
+                                device='cuda')
+        if rank == 0:
+            mmcv.mkdir_or_exist('.dist_test')
+            tmpdir = tempfile.mkdtemp(dir='.dist_test')
+            tmpdir = torch.tensor(
+                bytearray(tmpdir.encode()), dtype=torch.uint8, device='cuda')
+            dir_tensor[:len(tmpdir)] = tmpdir
+        dist.broadcast(dir_tensor, 0)
+        tmpdir = dir_tensor.cpu().numpy().tobytes().decode().rstrip()
+    else:
+        mmcv.mkdir_or_exist(tmpdir)
+    # dump the part result to the dir
+    mmcv.dump(result_part, osp.join(tmpdir, f'part_{rank}.pkl'))
+    dist.barrier()
+    # collect all parts
+    if rank != 0:
+        return None
+    else:
+        # load results of all parts from tmp dir
+        part_list = []
+        for i in range(world_size):
+            part_file = osp.join(tmpdir, f'part_{i}.pkl')
+            part_result = mmcv.load(part_file)
+            # When data is severely insufficient, an empty part_result
+            # on a certain gpu could makes the overall outputs empty.
+            if part_result:
+                part_list.append(part_result)
+        # sort the results
+        ordered_results = []
+        for res in zip(*part_list):
+            ordered_results.extend(list(res))
+        # the dataloader may pad some samples
+        ordered_results = ordered_results[:size]
+        # remove tmp dir
+        shutil.rmtree(tmpdir)
+        return ordered_results
+
+
+def collect_results_gpu(result_part, size):
+    """Collect results under gpu mode.
+
+    On gpu mode, this function will encode results to gpu tensors and use gpu
+    communication for results collection.
+
+    Args:
+        result_part (list): Result list containing result parts
+            to be collected.
+        size (int): Size of the results, commonly equal to length of
+            the results.
+
+    Returns:
+        list: The collected results.
+    """
+    rank, world_size = get_dist_info()
+    # dump result part to tensor with pickle
+    part_tensor = torch.tensor(
+        bytearray(pickle.dumps(result_part)), dtype=torch.uint8, device='cuda')
+    # gather all result part tensor shape
+    shape_tensor = torch.tensor(part_tensor.shape, device='cuda')
+    shape_list = [shape_tensor.clone() for _ in range(world_size)]
+    dist.all_gather(shape_list, shape_tensor)
+    # padding result part tensor to max length
+    shape_max = torch.tensor(shape_list).max()
+    part_send = torch.zeros(shape_max, dtype=torch.uint8, device='cuda')
+    part_send[:shape_tensor[0]] = part_tensor
+    part_recv_list = [
+        part_tensor.new_zeros(shape_max) for _ in range(world_size)
+    ]
+    # gather all result part
+    dist.all_gather(part_recv_list, part_send)
+
+    if rank == 0:
+        part_list = []
+        for recv, shape in zip(part_recv_list, shape_list):
+            part_result = pickle.loads(recv[:shape[0]].cpu().numpy().tobytes())
+            # When data is severely insufficient, an empty part_result
+            # on a certain gpu could makes the overall outputs empty.
+            if part_result:
+                part_list.append(part_result)
+        # sort the results
+        ordered_results = []
+        for res in zip(*part_list):
+            ordered_results.extend(list(res))
+        # the dataloader may pad some samples
+        ordered_results = ordered_results[:size]
+        return ordered_results
diff --git a/annotator/uniformer/mmcv/fileio/__init__.py b/annotator/uniformer/mmcv/fileio/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..2051b85f7e59bff7bdbaa131849ce8cd31f059a4
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/__init__.py
@@ -0,0 +1,11 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .file_client import BaseStorageBackend, FileClient
+from .handlers import BaseFileHandler, JsonHandler, PickleHandler, YamlHandler
+from .io import dump, load, register_handler
+from .parse import dict_from_file, list_from_file
+
+__all__ = [
+    'BaseStorageBackend', 'FileClient', 'load', 'dump', 'register_handler',
+    'BaseFileHandler', 'JsonHandler', 'PickleHandler', 'YamlHandler',
+    'list_from_file', 'dict_from_file'
+]
diff --git a/annotator/uniformer/mmcv/fileio/file_client.py b/annotator/uniformer/mmcv/fileio/file_client.py
new file mode 100644
index 0000000000000000000000000000000000000000..950f0c1aeab14b8e308a7455ccd64a95b5d98add
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/file_client.py
@@ -0,0 +1,1148 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import inspect
+import os
+import os.path as osp
+import re
+import tempfile
+import warnings
+from abc import ABCMeta, abstractmethod
+from contextlib import contextmanager
+from pathlib import Path
+from typing import Iterable, Iterator, Optional, Tuple, Union
+from urllib.request import urlopen
+
+import annotator.uniformer.mmcv as mmcv
+from annotator.uniformer.mmcv.utils.misc import has_method
+from annotator.uniformer.mmcv.utils.path import is_filepath
+
+
+class BaseStorageBackend(metaclass=ABCMeta):
+    """Abstract class of storage backends.
+
+    All backends need to implement two apis: ``get()`` and ``get_text()``.
+    ``get()`` reads the file as a byte stream and ``get_text()`` reads the file
+    as texts.
+    """
+
+    # a flag to indicate whether the backend can create a symlink for a file
+    _allow_symlink = False
+
+    @property
+    def name(self):
+        return self.__class__.__name__
+
+    @property
+    def allow_symlink(self):
+        return self._allow_symlink
+
+    @abstractmethod
+    def get(self, filepath):
+        pass
+
+    @abstractmethod
+    def get_text(self, filepath):
+        pass
+
+
+class CephBackend(BaseStorageBackend):
+    """Ceph storage backend (for internal use).
+
+    Args:
+        path_mapping (dict|None): path mapping dict from local path to Petrel
+            path. When ``path_mapping={'src': 'dst'}``, ``src`` in ``filepath``
+            will be replaced by ``dst``. Default: None.
+
+    .. warning::
+        :class:`mmcv.fileio.file_client.CephBackend` will be deprecated,
+        please use :class:`mmcv.fileio.file_client.PetrelBackend` instead.
+    """
+
+    def __init__(self, path_mapping=None):
+        try:
+            import ceph
+        except ImportError:
+            raise ImportError('Please install ceph to enable CephBackend.')
+
+        warnings.warn(
+            'CephBackend will be deprecated, please use PetrelBackend instead')
+        self._client = ceph.S3Client()
+        assert isinstance(path_mapping, dict) or path_mapping is None
+        self.path_mapping = path_mapping
+
+    def get(self, filepath):
+        filepath = str(filepath)
+        if self.path_mapping is not None:
+            for k, v in self.path_mapping.items():
+                filepath = filepath.replace(k, v)
+        value = self._client.Get(filepath)
+        value_buf = memoryview(value)
+        return value_buf
+
+    def get_text(self, filepath, encoding=None):
+        raise NotImplementedError
+
+
+class PetrelBackend(BaseStorageBackend):
+    """Petrel storage backend (for internal use).
+
+    PetrelBackend supports reading and writing data to multiple clusters.
+    If the file path contains the cluster name, PetrelBackend will read data
+    from specified cluster or write data to it. Otherwise, PetrelBackend will
+    access the default cluster.
+
+    Args:
+        path_mapping (dict, optional): Path mapping dict from local path to
+            Petrel path. When ``path_mapping={'src': 'dst'}``, ``src`` in
+            ``filepath`` will be replaced by ``dst``. Default: None.
+        enable_mc (bool, optional): Whether to enable memcached support.
+            Default: True.
+
+    Examples:
+        >>> filepath1 = 's3://path/of/file'
+        >>> filepath2 = 'cluster-name:s3://path/of/file'
+        >>> client = PetrelBackend()
+        >>> client.get(filepath1)  # get data from default cluster
+        >>> client.get(filepath2)  # get data from 'cluster-name' cluster
+    """
+
+    def __init__(self,
+                 path_mapping: Optional[dict] = None,
+                 enable_mc: bool = True):
+        try:
+            from petrel_client import client
+        except ImportError:
+            raise ImportError('Please install petrel_client to enable '
+                              'PetrelBackend.')
+
+        self._client = client.Client(enable_mc=enable_mc)
+        assert isinstance(path_mapping, dict) or path_mapping is None
+        self.path_mapping = path_mapping
+
+    def _map_path(self, filepath: Union[str, Path]) -> str:
+        """Map ``filepath`` to a string path whose prefix will be replaced by
+        :attr:`self.path_mapping`.
+
+        Args:
+            filepath (str): Path to be mapped.
+        """
+        filepath = str(filepath)
+        if self.path_mapping is not None:
+            for k, v in self.path_mapping.items():
+                filepath = filepath.replace(k, v)
+        return filepath
+
+    def _format_path(self, filepath: str) -> str:
+        """Convert a ``filepath`` to standard format of petrel oss.
+
+        If the ``filepath`` is concatenated by ``os.path.join``, in a Windows
+        environment, the ``filepath`` will be the format of
+        's3://bucket_name\\image.jpg'. By invoking :meth:`_format_path`, the
+        above ``filepath`` will be converted to 's3://bucket_name/image.jpg'.
+
+        Args:
+            filepath (str): Path to be formatted.
+        """
+        return re.sub(r'\\+', '/', filepath)
+
+    def get(self, filepath: Union[str, Path]) -> memoryview:
+        """Read data from a given ``filepath`` with 'rb' mode.
+
+        Args:
+            filepath (str or Path): Path to read data.
+
+        Returns:
+            memoryview: A memory view of expected bytes object to avoid
+                copying. The memoryview object can be converted to bytes by
+                ``value_buf.tobytes()``.
+        """
+        filepath = self._map_path(filepath)
+        filepath = self._format_path(filepath)
+        value = self._client.Get(filepath)
+        value_buf = memoryview(value)
+        return value_buf
+
+    def get_text(self,
+                 filepath: Union[str, Path],
+                 encoding: str = 'utf-8') -> str:
+        """Read data from a given ``filepath`` with 'r' mode.
+
+        Args:
+            filepath (str or Path): Path to read data.
+            encoding (str): The encoding format used to open the ``filepath``.
+                Default: 'utf-8'.
+
+        Returns:
+            str: Expected text reading from ``filepath``.
+        """
+        return str(self.get(filepath), encoding=encoding)
+
+    def put(self, obj: bytes, filepath: Union[str, Path]) -> None:
+        """Save data to a given ``filepath``.
+
+        Args:
+            obj (bytes): Data to be saved.
+            filepath (str or Path): Path to write data.
+        """
+        filepath = self._map_path(filepath)
+        filepath = self._format_path(filepath)
+        self._client.put(filepath, obj)
+
+    def put_text(self,
+                 obj: str,
+                 filepath: Union[str, Path],
+                 encoding: str = 'utf-8') -> None:
+        """Save data to a given ``filepath``.
+
+        Args:
+            obj (str): Data to be written.
+            filepath (str or Path): Path to write data.
+            encoding (str): The encoding format used to encode the ``obj``.
+                Default: 'utf-8'.
+        """
+        self.put(bytes(obj, encoding=encoding), filepath)
+
+    def remove(self, filepath: Union[str, Path]) -> None:
+        """Remove a file.
+
+        Args:
+            filepath (str or Path): Path to be removed.
+        """
+        if not has_method(self._client, 'delete'):
+            raise NotImplementedError(
+                ('Current version of Petrel Python SDK has not supported '
+                 'the `delete` method, please use a higher version or dev'
+                 ' branch instead.'))
+
+        filepath = self._map_path(filepath)
+        filepath = self._format_path(filepath)
+        self._client.delete(filepath)
+
+    def exists(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path exists.
+
+        Args:
+            filepath (str or Path): Path to be checked whether exists.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` exists, ``False`` otherwise.
+        """
+        if not (has_method(self._client, 'contains')
+                and has_method(self._client, 'isdir')):
+            raise NotImplementedError(
+                ('Current version of Petrel Python SDK has not supported '
+                 'the `contains` and `isdir` methods, please use a higher'
+                 'version or dev branch instead.'))
+
+        filepath = self._map_path(filepath)
+        filepath = self._format_path(filepath)
+        return self._client.contains(filepath) or self._client.isdir(filepath)
+
+    def isdir(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path is a directory.
+
+        Args:
+            filepath (str or Path): Path to be checked whether it is a
+                directory.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` points to a directory,
+                ``False`` otherwise.
+        """
+        if not has_method(self._client, 'isdir'):
+            raise NotImplementedError(
+                ('Current version of Petrel Python SDK has not supported '
+                 'the `isdir` method, please use a higher version or dev'
+                 ' branch instead.'))
+
+        filepath = self._map_path(filepath)
+        filepath = self._format_path(filepath)
+        return self._client.isdir(filepath)
+
+    def isfile(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path is a file.
+
+        Args:
+            filepath (str or Path): Path to be checked whether it is a file.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` points to a file, ``False``
+                otherwise.
+        """
+        if not has_method(self._client, 'contains'):
+            raise NotImplementedError(
+                ('Current version of Petrel Python SDK has not supported '
+                 'the `contains` method, please use a higher version or '
+                 'dev branch instead.'))
+
+        filepath = self._map_path(filepath)
+        filepath = self._format_path(filepath)
+        return self._client.contains(filepath)
+
+    def join_path(self, filepath: Union[str, Path],
+                  *filepaths: Union[str, Path]) -> str:
+        """Concatenate all file paths.
+
+        Args:
+            filepath (str or Path): Path to be concatenated.
+
+        Returns:
+            str: The result after concatenation.
+        """
+        filepath = self._format_path(self._map_path(filepath))
+        if filepath.endswith('/'):
+            filepath = filepath[:-1]
+        formatted_paths = [filepath]
+        for path in filepaths:
+            formatted_paths.append(self._format_path(self._map_path(path)))
+        return '/'.join(formatted_paths)
+
+    @contextmanager
+    def get_local_path(self, filepath: Union[str, Path]) -> Iterable[str]:
+        """Download a file from ``filepath`` and return a temporary path.
+
+        ``get_local_path`` is decorated by :meth:`contxtlib.contextmanager`. It
+        can be called with ``with`` statement, and when exists from the
+        ``with`` statement, the temporary path will be released.
+
+        Args:
+            filepath (str | Path): Download a file from ``filepath``.
+
+        Examples:
+            >>> client = PetrelBackend()
+            >>> # After existing from the ``with`` clause,
+            >>> # the path will be removed
+            >>> with client.get_local_path('s3://path/of/your/file') as path:
+            ...     # do something here
+
+        Yields:
+            Iterable[str]: Only yield one temporary path.
+        """
+        filepath = self._map_path(filepath)
+        filepath = self._format_path(filepath)
+        assert self.isfile(filepath)
+        try:
+            f = tempfile.NamedTemporaryFile(delete=False)
+            f.write(self.get(filepath))
+            f.close()
+            yield f.name
+        finally:
+            os.remove(f.name)
+
+    def list_dir_or_file(self,
+                         dir_path: Union[str, Path],
+                         list_dir: bool = True,
+                         list_file: bool = True,
+                         suffix: Optional[Union[str, Tuple[str]]] = None,
+                         recursive: bool = False) -> Iterator[str]:
+        """Scan a directory to find the interested directories or files in
+        arbitrary order.
+
+        Note:
+            Petrel has no concept of directories but it simulates the directory
+            hierarchy in the filesystem through public prefixes. In addition,
+            if the returned path ends with '/', it means the path is a public
+            prefix which is a logical directory.
+
+        Note:
+            :meth:`list_dir_or_file` returns the path relative to ``dir_path``.
+            In addition, the returned path of directory will not contains the
+            suffix '/' which is consistent with other backends.
+
+        Args:
+            dir_path (str | Path): Path of the directory.
+            list_dir (bool): List the directories. Default: True.
+            list_file (bool): List the path of files. Default: True.
+            suffix (str or tuple[str], optional):  File suffix
+                that we are interested in. Default: None.
+            recursive (bool): If set to True, recursively scan the
+                directory. Default: False.
+
+        Yields:
+            Iterable[str]: A relative path to ``dir_path``.
+        """
+        if not has_method(self._client, 'list'):
+            raise NotImplementedError(
+                ('Current version of Petrel Python SDK has not supported '
+                 'the `list` method, please use a higher version or dev'
+                 ' branch instead.'))
+
+        dir_path = self._map_path(dir_path)
+        dir_path = self._format_path(dir_path)
+        if list_dir and suffix is not None:
+            raise TypeError(
+                '`list_dir` should be False when `suffix` is not None')
+
+        if (suffix is not None) and not isinstance(suffix, (str, tuple)):
+            raise TypeError('`suffix` must be a string or tuple of strings')
+
+        # Petrel's simulated directory hierarchy assumes that directory paths
+        # should end with `/`
+        if not dir_path.endswith('/'):
+            dir_path += '/'
+
+        root = dir_path
+
+        def _list_dir_or_file(dir_path, list_dir, list_file, suffix,
+                              recursive):
+            for path in self._client.list(dir_path):
+                # the `self.isdir` is not used here to determine whether path
+                # is a directory, because `self.isdir` relies on
+                # `self._client.list`
+                if path.endswith('/'):  # a directory path
+                    next_dir_path = self.join_path(dir_path, path)
+                    if list_dir:
+                        # get the relative path and exclude the last
+                        # character '/'
+                        rel_dir = next_dir_path[len(root):-1]
+                        yield rel_dir
+                    if recursive:
+                        yield from _list_dir_or_file(next_dir_path, list_dir,
+                                                     list_file, suffix,
+                                                     recursive)
+                else:  # a file path
+                    absolute_path = self.join_path(dir_path, path)
+                    rel_path = absolute_path[len(root):]
+                    if (suffix is None
+                            or rel_path.endswith(suffix)) and list_file:
+                        yield rel_path
+
+        return _list_dir_or_file(dir_path, list_dir, list_file, suffix,
+                                 recursive)
+
+
+class MemcachedBackend(BaseStorageBackend):
+    """Memcached storage backend.
+
+    Attributes:
+        server_list_cfg (str): Config file for memcached server list.
+        client_cfg (str): Config file for memcached client.
+        sys_path (str | None): Additional path to be appended to `sys.path`.
+            Default: None.
+    """
+
+    def __init__(self, server_list_cfg, client_cfg, sys_path=None):
+        if sys_path is not None:
+            import sys
+            sys.path.append(sys_path)
+        try:
+            import mc
+        except ImportError:
+            raise ImportError(
+                'Please install memcached to enable MemcachedBackend.')
+
+        self.server_list_cfg = server_list_cfg
+        self.client_cfg = client_cfg
+        self._client = mc.MemcachedClient.GetInstance(self.server_list_cfg,
+                                                      self.client_cfg)
+        # mc.pyvector servers as a point which points to a memory cache
+        self._mc_buffer = mc.pyvector()
+
+    def get(self, filepath):
+        filepath = str(filepath)
+        import mc
+        self._client.Get(filepath, self._mc_buffer)
+        value_buf = mc.ConvertBuffer(self._mc_buffer)
+        return value_buf
+
+    def get_text(self, filepath, encoding=None):
+        raise NotImplementedError
+
+
+class LmdbBackend(BaseStorageBackend):
+    """Lmdb storage backend.
+
+    Args:
+        db_path (str): Lmdb database path.
+        readonly (bool, optional): Lmdb environment parameter. If True,
+            disallow any write operations. Default: True.
+        lock (bool, optional): Lmdb environment parameter. If False, when
+            concurrent access occurs, do not lock the database. Default: False.
+        readahead (bool, optional): Lmdb environment parameter. If False,
+            disable the OS filesystem readahead mechanism, which may improve
+            random read performance when a database is larger than RAM.
+            Default: False.
+
+    Attributes:
+        db_path (str): Lmdb database path.
+    """
+
+    def __init__(self,
+                 db_path,
+                 readonly=True,
+                 lock=False,
+                 readahead=False,
+                 **kwargs):
+        try:
+            import lmdb
+        except ImportError:
+            raise ImportError('Please install lmdb to enable LmdbBackend.')
+
+        self.db_path = str(db_path)
+        self._client = lmdb.open(
+            self.db_path,
+            readonly=readonly,
+            lock=lock,
+            readahead=readahead,
+            **kwargs)
+
+    def get(self, filepath):
+        """Get values according to the filepath.
+
+        Args:
+            filepath (str | obj:`Path`): Here, filepath is the lmdb key.
+        """
+        filepath = str(filepath)
+        with self._client.begin(write=False) as txn:
+            value_buf = txn.get(filepath.encode('ascii'))
+        return value_buf
+
+    def get_text(self, filepath, encoding=None):
+        raise NotImplementedError
+
+
+class HardDiskBackend(BaseStorageBackend):
+    """Raw hard disks storage backend."""
+
+    _allow_symlink = True
+
+    def get(self, filepath: Union[str, Path]) -> bytes:
+        """Read data from a given ``filepath`` with 'rb' mode.
+
+        Args:
+            filepath (str or Path): Path to read data.
+
+        Returns:
+            bytes: Expected bytes object.
+        """
+        with open(filepath, 'rb') as f:
+            value_buf = f.read()
+        return value_buf
+
+    def get_text(self,
+                 filepath: Union[str, Path],
+                 encoding: str = 'utf-8') -> str:
+        """Read data from a given ``filepath`` with 'r' mode.
+
+        Args:
+            filepath (str or Path): Path to read data.
+            encoding (str): The encoding format used to open the ``filepath``.
+                Default: 'utf-8'.
+
+        Returns:
+            str: Expected text reading from ``filepath``.
+        """
+        with open(filepath, 'r', encoding=encoding) as f:
+            value_buf = f.read()
+        return value_buf
+
+    def put(self, obj: bytes, filepath: Union[str, Path]) -> None:
+        """Write data to a given ``filepath`` with 'wb' mode.
+
+        Note:
+            ``put`` will create a directory if the directory of ``filepath``
+            does not exist.
+
+        Args:
+            obj (bytes): Data to be written.
+            filepath (str or Path): Path to write data.
+        """
+        mmcv.mkdir_or_exist(osp.dirname(filepath))
+        with open(filepath, 'wb') as f:
+            f.write(obj)
+
+    def put_text(self,
+                 obj: str,
+                 filepath: Union[str, Path],
+                 encoding: str = 'utf-8') -> None:
+        """Write data to a given ``filepath`` with 'w' mode.
+
+        Note:
+            ``put_text`` will create a directory if the directory of
+            ``filepath`` does not exist.
+
+        Args:
+            obj (str): Data to be written.
+            filepath (str or Path): Path to write data.
+            encoding (str): The encoding format used to open the ``filepath``.
+                Default: 'utf-8'.
+        """
+        mmcv.mkdir_or_exist(osp.dirname(filepath))
+        with open(filepath, 'w', encoding=encoding) as f:
+            f.write(obj)
+
+    def remove(self, filepath: Union[str, Path]) -> None:
+        """Remove a file.
+
+        Args:
+            filepath (str or Path): Path to be removed.
+        """
+        os.remove(filepath)
+
+    def exists(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path exists.
+
+        Args:
+            filepath (str or Path): Path to be checked whether exists.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` exists, ``False`` otherwise.
+        """
+        return osp.exists(filepath)
+
+    def isdir(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path is a directory.
+
+        Args:
+            filepath (str or Path): Path to be checked whether it is a
+                directory.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` points to a directory,
+                ``False`` otherwise.
+        """
+        return osp.isdir(filepath)
+
+    def isfile(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path is a file.
+
+        Args:
+            filepath (str or Path): Path to be checked whether it is a file.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` points to a file, ``False``
+                otherwise.
+        """
+        return osp.isfile(filepath)
+
+    def join_path(self, filepath: Union[str, Path],
+                  *filepaths: Union[str, Path]) -> str:
+        """Concatenate all file paths.
+
+        Join one or more filepath components intelligently. The return value
+        is the concatenation of filepath and any members of *filepaths.
+
+        Args:
+            filepath (str or Path): Path to be concatenated.
+
+        Returns:
+            str: The result of concatenation.
+        """
+        return osp.join(filepath, *filepaths)
+
+    @contextmanager
+    def get_local_path(
+            self, filepath: Union[str, Path]) -> Iterable[Union[str, Path]]:
+        """Only for unified API and do nothing."""
+        yield filepath
+
+    def list_dir_or_file(self,
+                         dir_path: Union[str, Path],
+                         list_dir: bool = True,
+                         list_file: bool = True,
+                         suffix: Optional[Union[str, Tuple[str]]] = None,
+                         recursive: bool = False) -> Iterator[str]:
+        """Scan a directory to find the interested directories or files in
+        arbitrary order.
+
+        Note:
+            :meth:`list_dir_or_file` returns the path relative to ``dir_path``.
+
+        Args:
+            dir_path (str | Path): Path of the directory.
+            list_dir (bool): List the directories. Default: True.
+            list_file (bool): List the path of files. Default: True.
+            suffix (str or tuple[str], optional):  File suffix
+                that we are interested in. Default: None.
+            recursive (bool): If set to True, recursively scan the
+                directory. Default: False.
+
+        Yields:
+            Iterable[str]: A relative path to ``dir_path``.
+        """
+        if list_dir and suffix is not None:
+            raise TypeError('`suffix` should be None when `list_dir` is True')
+
+        if (suffix is not None) and not isinstance(suffix, (str, tuple)):
+            raise TypeError('`suffix` must be a string or tuple of strings')
+
+        root = dir_path
+
+        def _list_dir_or_file(dir_path, list_dir, list_file, suffix,
+                              recursive):
+            for entry in os.scandir(dir_path):
+                if not entry.name.startswith('.') and entry.is_file():
+                    rel_path = osp.relpath(entry.path, root)
+                    if (suffix is None
+                            or rel_path.endswith(suffix)) and list_file:
+                        yield rel_path
+                elif osp.isdir(entry.path):
+                    if list_dir:
+                        rel_dir = osp.relpath(entry.path, root)
+                        yield rel_dir
+                    if recursive:
+                        yield from _list_dir_or_file(entry.path, list_dir,
+                                                     list_file, suffix,
+                                                     recursive)
+
+        return _list_dir_or_file(dir_path, list_dir, list_file, suffix,
+                                 recursive)
+
+
+class HTTPBackend(BaseStorageBackend):
+    """HTTP and HTTPS storage bachend."""
+
+    def get(self, filepath):
+        value_buf = urlopen(filepath).read()
+        return value_buf
+
+    def get_text(self, filepath, encoding='utf-8'):
+        value_buf = urlopen(filepath).read()
+        return value_buf.decode(encoding)
+
+    @contextmanager
+    def get_local_path(self, filepath: str) -> Iterable[str]:
+        """Download a file from ``filepath``.
+
+        ``get_local_path`` is decorated by :meth:`contxtlib.contextmanager`. It
+        can be called with ``with`` statement, and when exists from the
+        ``with`` statement, the temporary path will be released.
+
+        Args:
+            filepath (str): Download a file from ``filepath``.
+
+        Examples:
+            >>> client = HTTPBackend()
+            >>> # After existing from the ``with`` clause,
+            >>> # the path will be removed
+            >>> with client.get_local_path('http://path/of/your/file') as path:
+            ...     # do something here
+        """
+        try:
+            f = tempfile.NamedTemporaryFile(delete=False)
+            f.write(self.get(filepath))
+            f.close()
+            yield f.name
+        finally:
+            os.remove(f.name)
+
+
+class FileClient:
+    """A general file client to access files in different backends.
+
+    The client loads a file or text in a specified backend from its path
+    and returns it as a binary or text file. There are two ways to choose a
+    backend, the name of backend and the prefix of path. Although both of them
+    can be used to choose a storage backend, ``backend`` has a higher priority
+    that is if they are all set, the storage backend will be chosen by the
+    backend argument. If they are all `None`, the disk backend will be chosen.
+    Note that It can also register other backend accessor with a given name,
+    prefixes, and backend class. In addition, We use the singleton pattern to
+    avoid repeated object creation. If the arguments are the same, the same
+    object will be returned.
+
+    Args:
+        backend (str, optional): The storage backend type. Options are "disk",
+            "ceph", "memcached", "lmdb", "http" and "petrel". Default: None.
+        prefix (str, optional): The prefix of the registered storage backend.
+            Options are "s3", "http", "https". Default: None.
+
+    Examples:
+        >>> # only set backend
+        >>> file_client = FileClient(backend='petrel')
+        >>> # only set prefix
+        >>> file_client = FileClient(prefix='s3')
+        >>> # set both backend and prefix but use backend to choose client
+        >>> file_client = FileClient(backend='petrel', prefix='s3')
+        >>> # if the arguments are the same, the same object is returned
+        >>> file_client1 = FileClient(backend='petrel')
+        >>> file_client1 is file_client
+        True
+
+    Attributes:
+        client (:obj:`BaseStorageBackend`): The backend object.
+    """
+
+    _backends = {
+        'disk': HardDiskBackend,
+        'ceph': CephBackend,
+        'memcached': MemcachedBackend,
+        'lmdb': LmdbBackend,
+        'petrel': PetrelBackend,
+        'http': HTTPBackend,
+    }
+    # This collection is used to record the overridden backends, and when a
+    # backend appears in the collection, the singleton pattern is disabled for
+    # that backend, because if the singleton pattern is used, then the object
+    # returned will be the backend before overwriting
+    _overridden_backends = set()
+    _prefix_to_backends = {
+        's3': PetrelBackend,
+        'http': HTTPBackend,
+        'https': HTTPBackend,
+    }
+    _overridden_prefixes = set()
+
+    _instances = {}
+
+    def __new__(cls, backend=None, prefix=None, **kwargs):
+        if backend is None and prefix is None:
+            backend = 'disk'
+        if backend is not None and backend not in cls._backends:
+            raise ValueError(
+                f'Backend {backend} is not supported. Currently supported ones'
+                f' are {list(cls._backends.keys())}')
+        if prefix is not None and prefix not in cls._prefix_to_backends:
+            raise ValueError(
+                f'prefix {prefix} is not supported. Currently supported ones '
+                f'are {list(cls._prefix_to_backends.keys())}')
+
+        # concatenate the arguments to a unique key for determining whether
+        # objects with the same arguments were created
+        arg_key = f'{backend}:{prefix}'
+        for key, value in kwargs.items():
+            arg_key += f':{key}:{value}'
+
+        # if a backend was overridden, it will create a new object
+        if (arg_key in cls._instances
+                and backend not in cls._overridden_backends
+                and prefix not in cls._overridden_prefixes):
+            _instance = cls._instances[arg_key]
+        else:
+            # create a new object and put it to _instance
+            _instance = super().__new__(cls)
+            if backend is not None:
+                _instance.client = cls._backends[backend](**kwargs)
+            else:
+                _instance.client = cls._prefix_to_backends[prefix](**kwargs)
+
+            cls._instances[arg_key] = _instance
+
+        return _instance
+
+    @property
+    def name(self):
+        return self.client.name
+
+    @property
+    def allow_symlink(self):
+        return self.client.allow_symlink
+
+    @staticmethod
+    def parse_uri_prefix(uri: Union[str, Path]) -> Optional[str]:
+        """Parse the prefix of a uri.
+
+        Args:
+            uri (str | Path): Uri to be parsed that contains the file prefix.
+
+        Examples:
+            >>> FileClient.parse_uri_prefix('s3://path/of/your/file')
+            's3'
+
+        Returns:
+            str | None: Return the prefix of uri if the uri contains '://'
+                else ``None``.
+        """
+        assert is_filepath(uri)
+        uri = str(uri)
+        if '://' not in uri:
+            return None
+        else:
+            prefix, _ = uri.split('://')
+            # In the case of PetrelBackend, the prefix may contains the cluster
+            # name like clusterName:s3
+            if ':' in prefix:
+                _, prefix = prefix.split(':')
+            return prefix
+
+    @classmethod
+    def infer_client(cls,
+                     file_client_args: Optional[dict] = None,
+                     uri: Optional[Union[str, Path]] = None) -> 'FileClient':
+        """Infer a suitable file client based on the URI and arguments.
+
+        Args:
+            file_client_args (dict, optional): Arguments to instantiate a
+                FileClient. Default: None.
+            uri (str | Path, optional): Uri to be parsed that contains the file
+                prefix. Default: None.
+
+        Examples:
+            >>> uri = 's3://path/of/your/file'
+            >>> file_client = FileClient.infer_client(uri=uri)
+            >>> file_client_args = {'backend': 'petrel'}
+            >>> file_client = FileClient.infer_client(file_client_args)
+
+        Returns:
+            FileClient: Instantiated FileClient object.
+        """
+        assert file_client_args is not None or uri is not None
+        if file_client_args is None:
+            file_prefix = cls.parse_uri_prefix(uri)  # type: ignore
+            return cls(prefix=file_prefix)
+        else:
+            return cls(**file_client_args)
+
+    @classmethod
+    def _register_backend(cls, name, backend, force=False, prefixes=None):
+        if not isinstance(name, str):
+            raise TypeError('the backend name should be a string, '
+                            f'but got {type(name)}')
+        if not inspect.isclass(backend):
+            raise TypeError(
+                f'backend should be a class but got {type(backend)}')
+        if not issubclass(backend, BaseStorageBackend):
+            raise TypeError(
+                f'backend {backend} is not a subclass of BaseStorageBackend')
+        if not force and name in cls._backends:
+            raise KeyError(
+                f'{name} is already registered as a storage backend, '
+                'add "force=True" if you want to override it')
+
+        if name in cls._backends and force:
+            cls._overridden_backends.add(name)
+        cls._backends[name] = backend
+
+        if prefixes is not None:
+            if isinstance(prefixes, str):
+                prefixes = [prefixes]
+            else:
+                assert isinstance(prefixes, (list, tuple))
+            for prefix in prefixes:
+                if prefix not in cls._prefix_to_backends:
+                    cls._prefix_to_backends[prefix] = backend
+                elif (prefix in cls._prefix_to_backends) and force:
+                    cls._overridden_prefixes.add(prefix)
+                    cls._prefix_to_backends[prefix] = backend
+                else:
+                    raise KeyError(
+                        f'{prefix} is already registered as a storage backend,'
+                        ' add "force=True" if you want to override it')
+
+    @classmethod
+    def register_backend(cls, name, backend=None, force=False, prefixes=None):
+        """Register a backend to FileClient.
+
+        This method can be used as a normal class method or a decorator.
+
+        .. code-block:: python
+
+            class NewBackend(BaseStorageBackend):
+
+                def get(self, filepath):
+                    return filepath
+
+                def get_text(self, filepath):
+                    return filepath
+
+            FileClient.register_backend('new', NewBackend)
+
+        or
+
+        .. code-block:: python
+
+            @FileClient.register_backend('new')
+            class NewBackend(BaseStorageBackend):
+
+                def get(self, filepath):
+                    return filepath
+
+                def get_text(self, filepath):
+                    return filepath
+
+        Args:
+            name (str): The name of the registered backend.
+            backend (class, optional): The backend class to be registered,
+                which must be a subclass of :class:`BaseStorageBackend`.
+                When this method is used as a decorator, backend is None.
+                Defaults to None.
+            force (bool, optional): Whether to override the backend if the name
+                has already been registered. Defaults to False.
+            prefixes (str or list[str] or tuple[str], optional): The prefixes
+                of the registered storage backend. Default: None.
+                `New in version 1.3.15.`
+        """
+        if backend is not None:
+            cls._register_backend(
+                name, backend, force=force, prefixes=prefixes)
+            return
+
+        def _register(backend_cls):
+            cls._register_backend(
+                name, backend_cls, force=force, prefixes=prefixes)
+            return backend_cls
+
+        return _register
+
+    def get(self, filepath: Union[str, Path]) -> Union[bytes, memoryview]:
+        """Read data from a given ``filepath`` with 'rb' mode.
+
+        Note:
+            There are two types of return values for ``get``, one is ``bytes``
+            and the other is ``memoryview``. The advantage of using memoryview
+            is that you can avoid copying, and if you want to convert it to
+            ``bytes``, you can use ``.tobytes()``.
+
+        Args:
+            filepath (str or Path): Path to read data.
+
+        Returns:
+            bytes | memoryview: Expected bytes object or a memory view of the
+                bytes object.
+        """
+        return self.client.get(filepath)
+
+    def get_text(self, filepath: Union[str, Path], encoding='utf-8') -> str:
+        """Read data from a given ``filepath`` with 'r' mode.
+
+        Args:
+            filepath (str or Path): Path to read data.
+            encoding (str): The encoding format used to open the ``filepath``.
+                Default: 'utf-8'.
+
+        Returns:
+            str: Expected text reading from ``filepath``.
+        """
+        return self.client.get_text(filepath, encoding)
+
+    def put(self, obj: bytes, filepath: Union[str, Path]) -> None:
+        """Write data to a given ``filepath`` with 'wb' mode.
+
+        Note:
+            ``put`` should create a directory if the directory of ``filepath``
+            does not exist.
+
+        Args:
+            obj (bytes): Data to be written.
+            filepath (str or Path): Path to write data.
+        """
+        self.client.put(obj, filepath)
+
+    def put_text(self, obj: str, filepath: Union[str, Path]) -> None:
+        """Write data to a given ``filepath`` with 'w' mode.
+
+        Note:
+            ``put_text`` should create a directory if the directory of
+            ``filepath`` does not exist.
+
+        Args:
+            obj (str): Data to be written.
+            filepath (str or Path): Path to write data.
+            encoding (str, optional): The encoding format used to open the
+                `filepath`. Default: 'utf-8'.
+        """
+        self.client.put_text(obj, filepath)
+
+    def remove(self, filepath: Union[str, Path]) -> None:
+        """Remove a file.
+
+        Args:
+            filepath (str, Path): Path to be removed.
+        """
+        self.client.remove(filepath)
+
+    def exists(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path exists.
+
+        Args:
+            filepath (str or Path): Path to be checked whether exists.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` exists, ``False`` otherwise.
+        """
+        return self.client.exists(filepath)
+
+    def isdir(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path is a directory.
+
+        Args:
+            filepath (str or Path): Path to be checked whether it is a
+                directory.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` points to a directory,
+                ``False`` otherwise.
+        """
+        return self.client.isdir(filepath)
+
+    def isfile(self, filepath: Union[str, Path]) -> bool:
+        """Check whether a file path is a file.
+
+        Args:
+            filepath (str or Path): Path to be checked whether it is a file.
+
+        Returns:
+            bool: Return ``True`` if ``filepath`` points to a file, ``False``
+                otherwise.
+        """
+        return self.client.isfile(filepath)
+
+    def join_path(self, filepath: Union[str, Path],
+                  *filepaths: Union[str, Path]) -> str:
+        """Concatenate all file paths.
+
+        Join one or more filepath components intelligently. The return value
+        is the concatenation of filepath and any members of *filepaths.
+
+        Args:
+            filepath (str or Path): Path to be concatenated.
+
+        Returns:
+            str: The result of concatenation.
+        """
+        return self.client.join_path(filepath, *filepaths)
+
+    @contextmanager
+    def get_local_path(self, filepath: Union[str, Path]) -> Iterable[str]:
+        """Download data from ``filepath`` and write the data to local path.
+
+        ``get_local_path`` is decorated by :meth:`contxtlib.contextmanager`. It
+        can be called with ``with`` statement, and when exists from the
+        ``with`` statement, the temporary path will be released.
+
+        Note:
+            If the ``filepath`` is a local path, just return itself.
+
+        .. warning::
+            ``get_local_path`` is an experimental interface that may change in
+            the future.
+
+        Args:
+            filepath (str or Path): Path to be read data.
+
+        Examples:
+            >>> file_client = FileClient(prefix='s3')
+            >>> with file_client.get_local_path('s3://bucket/abc.jpg') as path:
+            ...     # do something here
+
+        Yields:
+            Iterable[str]: Only yield one path.
+        """
+        with self.client.get_local_path(str(filepath)) as local_path:
+            yield local_path
+
+    def list_dir_or_file(self,
+                         dir_path: Union[str, Path],
+                         list_dir: bool = True,
+                         list_file: bool = True,
+                         suffix: Optional[Union[str, Tuple[str]]] = None,
+                         recursive: bool = False) -> Iterator[str]:
+        """Scan a directory to find the interested directories or files in
+        arbitrary order.
+
+        Note:
+            :meth:`list_dir_or_file` returns the path relative to ``dir_path``.
+
+        Args:
+            dir_path (str | Path): Path of the directory.
+            list_dir (bool): List the directories. Default: True.
+            list_file (bool): List the path of files. Default: True.
+            suffix (str or tuple[str], optional):  File suffix
+                that we are interested in. Default: None.
+            recursive (bool): If set to True, recursively scan the
+                directory. Default: False.
+
+        Yields:
+            Iterable[str]: A relative path to ``dir_path``.
+        """
+        yield from self.client.list_dir_or_file(dir_path, list_dir, list_file,
+                                                suffix, recursive)
diff --git a/annotator/uniformer/mmcv/fileio/handlers/__init__.py b/annotator/uniformer/mmcv/fileio/handlers/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..aa24d91972837b8756b225f4879bac20436eb72a
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/handlers/__init__.py
@@ -0,0 +1,7 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .base import BaseFileHandler
+from .json_handler import JsonHandler
+from .pickle_handler import PickleHandler
+from .yaml_handler import YamlHandler
+
+__all__ = ['BaseFileHandler', 'JsonHandler', 'PickleHandler', 'YamlHandler']
diff --git a/annotator/uniformer/mmcv/fileio/handlers/base.py b/annotator/uniformer/mmcv/fileio/handlers/base.py
new file mode 100644
index 0000000000000000000000000000000000000000..288878bc57282fbb2f12b32290152ca8e9d3cab0
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/handlers/base.py
@@ -0,0 +1,30 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from abc import ABCMeta, abstractmethod
+
+
+class BaseFileHandler(metaclass=ABCMeta):
+    # `str_like` is a flag to indicate whether the type of file object is
+    # str-like object or bytes-like object. Pickle only processes bytes-like
+    # objects but json only processes str-like object. If it is str-like
+    # object, `StringIO` will be used to process the buffer.
+    str_like = True
+
+    @abstractmethod
+    def load_from_fileobj(self, file, **kwargs):
+        pass
+
+    @abstractmethod
+    def dump_to_fileobj(self, obj, file, **kwargs):
+        pass
+
+    @abstractmethod
+    def dump_to_str(self, obj, **kwargs):
+        pass
+
+    def load_from_path(self, filepath, mode='r', **kwargs):
+        with open(filepath, mode) as f:
+            return self.load_from_fileobj(f, **kwargs)
+
+    def dump_to_path(self, obj, filepath, mode='w', **kwargs):
+        with open(filepath, mode) as f:
+            self.dump_to_fileobj(obj, f, **kwargs)
diff --git a/annotator/uniformer/mmcv/fileio/handlers/json_handler.py b/annotator/uniformer/mmcv/fileio/handlers/json_handler.py
new file mode 100644
index 0000000000000000000000000000000000000000..18d4f15f74139d20adff18b20be5529c592a66b6
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/handlers/json_handler.py
@@ -0,0 +1,36 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import json
+
+import numpy as np
+
+from .base import BaseFileHandler
+
+
+def set_default(obj):
+    """Set default json values for non-serializable values.
+
+    It helps convert ``set``, ``range`` and ``np.ndarray`` data types to list.
+    It also converts ``np.generic`` (including ``np.int32``, ``np.float32``,
+    etc.) into plain numbers of plain python built-in types.
+    """
+    if isinstance(obj, (set, range)):
+        return list(obj)
+    elif isinstance(obj, np.ndarray):
+        return obj.tolist()
+    elif isinstance(obj, np.generic):
+        return obj.item()
+    raise TypeError(f'{type(obj)} is unsupported for json dump')
+
+
+class JsonHandler(BaseFileHandler):
+
+    def load_from_fileobj(self, file):
+        return json.load(file)
+
+    def dump_to_fileobj(self, obj, file, **kwargs):
+        kwargs.setdefault('default', set_default)
+        json.dump(obj, file, **kwargs)
+
+    def dump_to_str(self, obj, **kwargs):
+        kwargs.setdefault('default', set_default)
+        return json.dumps(obj, **kwargs)
diff --git a/annotator/uniformer/mmcv/fileio/handlers/pickle_handler.py b/annotator/uniformer/mmcv/fileio/handlers/pickle_handler.py
new file mode 100644
index 0000000000000000000000000000000000000000..b37c79bed4ef9fd8913715e62dbe3fc5cafdc3aa
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/handlers/pickle_handler.py
@@ -0,0 +1,28 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import pickle
+
+from .base import BaseFileHandler
+
+
+class PickleHandler(BaseFileHandler):
+
+    str_like = False
+
+    def load_from_fileobj(self, file, **kwargs):
+        return pickle.load(file, **kwargs)
+
+    def load_from_path(self, filepath, **kwargs):
+        return super(PickleHandler, self).load_from_path(
+            filepath, mode='rb', **kwargs)
+
+    def dump_to_str(self, obj, **kwargs):
+        kwargs.setdefault('protocol', 2)
+        return pickle.dumps(obj, **kwargs)
+
+    def dump_to_fileobj(self, obj, file, **kwargs):
+        kwargs.setdefault('protocol', 2)
+        pickle.dump(obj, file, **kwargs)
+
+    def dump_to_path(self, obj, filepath, **kwargs):
+        super(PickleHandler, self).dump_to_path(
+            obj, filepath, mode='wb', **kwargs)
diff --git a/annotator/uniformer/mmcv/fileio/handlers/yaml_handler.py b/annotator/uniformer/mmcv/fileio/handlers/yaml_handler.py
new file mode 100644
index 0000000000000000000000000000000000000000..c5aa2eea1e8c76f8baf753d1c8c959dee665e543
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/handlers/yaml_handler.py
@@ -0,0 +1,24 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import yaml
+
+try:
+    from yaml import CLoader as Loader, CDumper as Dumper
+except ImportError:
+    from yaml import Loader, Dumper
+
+from .base import BaseFileHandler  # isort:skip
+
+
+class YamlHandler(BaseFileHandler):
+
+    def load_from_fileobj(self, file, **kwargs):
+        kwargs.setdefault('Loader', Loader)
+        return yaml.load(file, **kwargs)
+
+    def dump_to_fileobj(self, obj, file, **kwargs):
+        kwargs.setdefault('Dumper', Dumper)
+        yaml.dump(obj, file, **kwargs)
+
+    def dump_to_str(self, obj, **kwargs):
+        kwargs.setdefault('Dumper', Dumper)
+        return yaml.dump(obj, **kwargs)
diff --git a/annotator/uniformer/mmcv/fileio/io.py b/annotator/uniformer/mmcv/fileio/io.py
new file mode 100644
index 0000000000000000000000000000000000000000..aaefde58aa3ea5b58f86249ce7e1c40c186eb8dd
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/io.py
@@ -0,0 +1,151 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from io import BytesIO, StringIO
+from pathlib import Path
+
+from ..utils import is_list_of, is_str
+from .file_client import FileClient
+from .handlers import BaseFileHandler, JsonHandler, PickleHandler, YamlHandler
+
+file_handlers = {
+    'json': JsonHandler(),
+    'yaml': YamlHandler(),
+    'yml': YamlHandler(),
+    'pickle': PickleHandler(),
+    'pkl': PickleHandler()
+}
+
+
+def load(file, file_format=None, file_client_args=None, **kwargs):
+    """Load data from json/yaml/pickle files.
+
+    This method provides a unified api for loading data from serialized files.
+
+    Note:
+        In v1.3.16 and later, ``load`` supports loading data from serialized
+        files those can be storaged in different backends.
+
+    Args:
+        file (str or :obj:`Path` or file-like object): Filename or a file-like
+            object.
+        file_format (str, optional): If not specified, the file format will be
+            inferred from the file extension, otherwise use the specified one.
+            Currently supported formats include "json", "yaml/yml" and
+            "pickle/pkl".
+        file_client_args (dict, optional): Arguments to instantiate a
+            FileClient. See :class:`mmcv.fileio.FileClient` for details.
+            Default: None.
+
+    Examples:
+        >>> load('/path/of/your/file')  # file is storaged in disk
+        >>> load('https://path/of/your/file')  # file is storaged in Internet
+        >>> load('s3://path/of/your/file')  # file is storaged in petrel
+
+    Returns:
+        The content from the file.
+    """
+    if isinstance(file, Path):
+        file = str(file)
+    if file_format is None and is_str(file):
+        file_format = file.split('.')[-1]
+    if file_format not in file_handlers:
+        raise TypeError(f'Unsupported format: {file_format}')
+
+    handler = file_handlers[file_format]
+    if is_str(file):
+        file_client = FileClient.infer_client(file_client_args, file)
+        if handler.str_like:
+            with StringIO(file_client.get_text(file)) as f:
+                obj = handler.load_from_fileobj(f, **kwargs)
+        else:
+            with BytesIO(file_client.get(file)) as f:
+                obj = handler.load_from_fileobj(f, **kwargs)
+    elif hasattr(file, 'read'):
+        obj = handler.load_from_fileobj(file, **kwargs)
+    else:
+        raise TypeError('"file" must be a filepath str or a file-object')
+    return obj
+
+
+def dump(obj, file=None, file_format=None, file_client_args=None, **kwargs):
+    """Dump data to json/yaml/pickle strings or files.
+
+    This method provides a unified api for dumping data as strings or to files,
+    and also supports custom arguments for each file format.
+
+    Note:
+        In v1.3.16 and later, ``dump`` supports dumping data as strings or to
+        files which is saved to different backends.
+
+    Args:
+        obj (any): The python object to be dumped.
+        file (str or :obj:`Path` or file-like object, optional): If not
+            specified, then the object is dumped to a str, otherwise to a file
+            specified by the filename or file-like object.
+        file_format (str, optional): Same as :func:`load`.
+        file_client_args (dict, optional): Arguments to instantiate a
+            FileClient. See :class:`mmcv.fileio.FileClient` for details.
+            Default: None.
+
+    Examples:
+        >>> dump('hello world', '/path/of/your/file')  # disk
+        >>> dump('hello world', 's3://path/of/your/file')  # ceph or petrel
+
+    Returns:
+        bool: True for success, False otherwise.
+    """
+    if isinstance(file, Path):
+        file = str(file)
+    if file_format is None:
+        if is_str(file):
+            file_format = file.split('.')[-1]
+        elif file is None:
+            raise ValueError(
+                'file_format must be specified since file is None')
+    if file_format not in file_handlers:
+        raise TypeError(f'Unsupported format: {file_format}')
+
+    handler = file_handlers[file_format]
+    if file is None:
+        return handler.dump_to_str(obj, **kwargs)
+    elif is_str(file):
+        file_client = FileClient.infer_client(file_client_args, file)
+        if handler.str_like:
+            with StringIO() as f:
+                handler.dump_to_fileobj(obj, f, **kwargs)
+                file_client.put_text(f.getvalue(), file)
+        else:
+            with BytesIO() as f:
+                handler.dump_to_fileobj(obj, f, **kwargs)
+                file_client.put(f.getvalue(), file)
+    elif hasattr(file, 'write'):
+        handler.dump_to_fileobj(obj, file, **kwargs)
+    else:
+        raise TypeError('"file" must be a filename str or a file-object')
+
+
+def _register_handler(handler, file_formats):
+    """Register a handler for some file extensions.
+
+    Args:
+        handler (:obj:`BaseFileHandler`): Handler to be registered.
+        file_formats (str or list[str]): File formats to be handled by this
+            handler.
+    """
+    if not isinstance(handler, BaseFileHandler):
+        raise TypeError(
+            f'handler must be a child of BaseFileHandler, not {type(handler)}')
+    if isinstance(file_formats, str):
+        file_formats = [file_formats]
+    if not is_list_of(file_formats, str):
+        raise TypeError('file_formats must be a str or a list of str')
+    for ext in file_formats:
+        file_handlers[ext] = handler
+
+
+def register_handler(file_formats, **kwargs):
+
+    def wrap(cls):
+        _register_handler(cls(**kwargs), file_formats)
+        return cls
+
+    return wrap
diff --git a/annotator/uniformer/mmcv/fileio/parse.py b/annotator/uniformer/mmcv/fileio/parse.py
new file mode 100644
index 0000000000000000000000000000000000000000..f60f0d611b8d75692221d0edd7dc993b0a6445c9
--- /dev/null
+++ b/annotator/uniformer/mmcv/fileio/parse.py
@@ -0,0 +1,97 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+
+from io import StringIO
+
+from .file_client import FileClient
+
+
+def list_from_file(filename,
+                   prefix='',
+                   offset=0,
+                   max_num=0,
+                   encoding='utf-8',
+                   file_client_args=None):
+    """Load a text file and parse the content as a list of strings.
+
+    Note:
+        In v1.3.16 and later, ``list_from_file`` supports loading a text file
+        which can be storaged in different backends and parsing the content as
+        a list for strings.
+
+    Args:
+        filename (str): Filename.
+        prefix (str): The prefix to be inserted to the beginning of each item.
+        offset (int): The offset of lines.
+        max_num (int): The maximum number of lines to be read,
+            zeros and negatives mean no limitation.
+        encoding (str): Encoding used to open the file. Default utf-8.
+        file_client_args (dict, optional): Arguments to instantiate a
+            FileClient. See :class:`mmcv.fileio.FileClient` for details.
+            Default: None.
+
+    Examples:
+        >>> list_from_file('/path/of/your/file')  # disk
+        ['hello', 'world']
+        >>> list_from_file('s3://path/of/your/file')  # ceph or petrel
+        ['hello', 'world']
+
+    Returns:
+        list[str]: A list of strings.
+    """
+    cnt = 0
+    item_list = []
+    file_client = FileClient.infer_client(file_client_args, filename)
+    with StringIO(file_client.get_text(filename, encoding)) as f:
+        for _ in range(offset):
+            f.readline()
+        for line in f:
+            if 0 < max_num <= cnt:
+                break
+            item_list.append(prefix + line.rstrip('\n\r'))
+            cnt += 1
+    return item_list
+
+
+def dict_from_file(filename,
+                   key_type=str,
+                   encoding='utf-8',
+                   file_client_args=None):
+    """Load a text file and parse the content as a dict.
+
+    Each line of the text file will be two or more columns split by
+    whitespaces or tabs. The first column will be parsed as dict keys, and
+    the following columns will be parsed as dict values.
+
+    Note:
+        In v1.3.16 and later, ``dict_from_file`` supports loading a text file
+        which can be storaged in different backends and parsing the content as
+        a dict.
+
+    Args:
+        filename(str): Filename.
+        key_type(type): Type of the dict keys. str is user by default and
+            type conversion will be performed if specified.
+        encoding (str): Encoding used to open the file. Default utf-8.
+        file_client_args (dict, optional): Arguments to instantiate a
+            FileClient. See :class:`mmcv.fileio.FileClient` for details.
+            Default: None.
+
+    Examples:
+        >>> dict_from_file('/path/of/your/file')  # disk
+        {'key1': 'value1', 'key2': 'value2'}
+        >>> dict_from_file('s3://path/of/your/file')  # ceph or petrel
+        {'key1': 'value1', 'key2': 'value2'}
+
+    Returns:
+        dict: The parsed contents.
+    """
+    mapping = {}
+    file_client = FileClient.infer_client(file_client_args, filename)
+    with StringIO(file_client.get_text(filename, encoding)) as f:
+        for line in f:
+            items = line.rstrip('\n').split()
+            assert len(items) >= 2
+            key = key_type(items[0])
+            val = items[1:] if len(items) > 2 else items[1]
+            mapping[key] = val
+    return mapping
diff --git a/annotator/uniformer/mmcv/image/__init__.py b/annotator/uniformer/mmcv/image/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..d0051d609d3de4e7562e3fe638335c66617c4d91
--- /dev/null
+++ b/annotator/uniformer/mmcv/image/__init__.py
@@ -0,0 +1,28 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .colorspace import (bgr2gray, bgr2hls, bgr2hsv, bgr2rgb, bgr2ycbcr,
+                         gray2bgr, gray2rgb, hls2bgr, hsv2bgr, imconvert,
+                         rgb2bgr, rgb2gray, rgb2ycbcr, ycbcr2bgr, ycbcr2rgb)
+from .geometric import (cutout, imcrop, imflip, imflip_, impad,
+                        impad_to_multiple, imrescale, imresize, imresize_like,
+                        imresize_to_multiple, imrotate, imshear, imtranslate,
+                        rescale_size)
+from .io import imfrombytes, imread, imwrite, supported_backends, use_backend
+from .misc import tensor2imgs
+from .photometric import (adjust_brightness, adjust_color, adjust_contrast,
+                          adjust_lighting, adjust_sharpness, auto_contrast,
+                          clahe, imdenormalize, imequalize, iminvert,
+                          imnormalize, imnormalize_, lut_transform, posterize,
+                          solarize)
+
+__all__ = [
+    'bgr2gray', 'bgr2hls', 'bgr2hsv', 'bgr2rgb', 'gray2bgr', 'gray2rgb',
+    'hls2bgr', 'hsv2bgr', 'imconvert', 'rgb2bgr', 'rgb2gray', 'imrescale',
+    'imresize', 'imresize_like', 'imresize_to_multiple', 'rescale_size',
+    'imcrop', 'imflip', 'imflip_', 'impad', 'impad_to_multiple', 'imrotate',
+    'imfrombytes', 'imread', 'imwrite', 'supported_backends', 'use_backend',
+    'imdenormalize', 'imnormalize', 'imnormalize_', 'iminvert', 'posterize',
+    'solarize', 'rgb2ycbcr', 'bgr2ycbcr', 'ycbcr2rgb', 'ycbcr2bgr',
+    'tensor2imgs', 'imshear', 'imtranslate', 'adjust_color', 'imequalize',
+    'adjust_brightness', 'adjust_contrast', 'lut_transform', 'clahe',
+    'adjust_sharpness', 'auto_contrast', 'cutout', 'adjust_lighting'
+]
diff --git a/annotator/uniformer/mmcv/image/colorspace.py b/annotator/uniformer/mmcv/image/colorspace.py
new file mode 100644
index 0000000000000000000000000000000000000000..814533952fdfda23d67cb6a3073692d8c1156add
--- /dev/null
+++ b/annotator/uniformer/mmcv/image/colorspace.py
@@ -0,0 +1,306 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import cv2
+import numpy as np
+
+
+def imconvert(img, src, dst):
+    """Convert an image from the src colorspace to dst colorspace.
+
+    Args:
+        img (ndarray): The input image.
+        src (str): The source colorspace, e.g., 'rgb', 'hsv'.
+        dst (str): The destination colorspace, e.g., 'rgb', 'hsv'.
+
+    Returns:
+        ndarray: The converted image.
+    """
+    code = getattr(cv2, f'COLOR_{src.upper()}2{dst.upper()}')
+    out_img = cv2.cvtColor(img, code)
+    return out_img
+
+
+def bgr2gray(img, keepdim=False):
+    """Convert a BGR image to grayscale image.
+
+    Args:
+        img (ndarray): The input image.
+        keepdim (bool): If False (by default), then return the grayscale image
+            with 2 dims, otherwise 3 dims.
+
+    Returns:
+        ndarray: The converted grayscale image.
+    """
+    out_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
+    if keepdim:
+        out_img = out_img[..., None]
+    return out_img
+
+
+def rgb2gray(img, keepdim=False):
+    """Convert a RGB image to grayscale image.
+
+    Args:
+        img (ndarray): The input image.
+        keepdim (bool): If False (by default), then return the grayscale image
+            with 2 dims, otherwise 3 dims.
+
+    Returns:
+        ndarray: The converted grayscale image.
+    """
+    out_img = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
+    if keepdim:
+        out_img = out_img[..., None]
+    return out_img
+
+
+def gray2bgr(img):
+    """Convert a grayscale image to BGR image.
+
+    Args:
+        img (ndarray): The input image.
+
+    Returns:
+        ndarray: The converted BGR image.
+    """
+    img = img[..., None] if img.ndim == 2 else img
+    out_img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
+    return out_img
+
+
+def gray2rgb(img):
+    """Convert a grayscale image to RGB image.
+
+    Args:
+        img (ndarray): The input image.
+
+    Returns:
+        ndarray: The converted RGB image.
+    """
+    img = img[..., None] if img.ndim == 2 else img
+    out_img = cv2.cvtColor(img, cv2.COLOR_GRAY2RGB)
+    return out_img
+
+
+def _convert_input_type_range(img):
+    """Convert the type and range of the input image.
+
+    It converts the input image to np.float32 type and range of [0, 1].
+    It is mainly used for pre-processing the input image in colorspace
+    conversion functions such as rgb2ycbcr and ycbcr2rgb.
+
+    Args:
+        img (ndarray): The input image. It accepts:
+            1. np.uint8 type with range [0, 255];
+            2. np.float32 type with range [0, 1].
+
+    Returns:
+        (ndarray): The converted image with type of np.float32 and range of
+            [0, 1].
+    """
+    img_type = img.dtype
+    img = img.astype(np.float32)
+    if img_type == np.float32:
+        pass
+    elif img_type == np.uint8:
+        img /= 255.
+    else:
+        raise TypeError('The img type should be np.float32 or np.uint8, '
+                        f'but got {img_type}')
+    return img
+
+
+def _convert_output_type_range(img, dst_type):
+    """Convert the type and range of the image according to dst_type.
+
+    It converts the image to desired type and range. If `dst_type` is np.uint8,
+    images will be converted to np.uint8 type with range [0, 255]. If
+    `dst_type` is np.float32, it converts the image to np.float32 type with
+    range [0, 1].
+    It is mainly used for post-processing images in colorspace conversion
+    functions such as rgb2ycbcr and ycbcr2rgb.
+
+    Args:
+        img (ndarray): The image to be converted with np.float32 type and
+            range [0, 255].
+        dst_type (np.uint8 | np.float32): If dst_type is np.uint8, it
+            converts the image to np.uint8 type with range [0, 255]. If
+            dst_type is np.float32, it converts the image to np.float32 type
+            with range [0, 1].
+
+    Returns:
+        (ndarray): The converted image with desired type and range.
+    """
+    if dst_type not in (np.uint8, np.float32):
+        raise TypeError('The dst_type should be np.float32 or np.uint8, '
+                        f'but got {dst_type}')
+    if dst_type == np.uint8:
+        img = img.round()
+    else:
+        img /= 255.
+    return img.astype(dst_type)
+
+
+def rgb2ycbcr(img, y_only=False):
+    """Convert a RGB image to YCbCr image.
+
+    This function produces the same results as Matlab's `rgb2ycbcr` function.
+    It implements the ITU-R BT.601 conversion for standard-definition
+    television. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#ITU-R_BT.601_conversion.
+
+    It differs from a similar function in cv2.cvtColor: `RGB <-> YCrCb`.
+    In OpenCV, it implements a JPEG conversion. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#JPEG_conversion.
+
+    Args:
+        img (ndarray): The input image. It accepts:
+            1. np.uint8 type with range [0, 255];
+            2. np.float32 type with range [0, 1].
+        y_only (bool): Whether to only return Y channel. Default: False.
+
+    Returns:
+        ndarray: The converted YCbCr image. The output image has the same type
+            and range as input image.
+    """
+    img_type = img.dtype
+    img = _convert_input_type_range(img)
+    if y_only:
+        out_img = np.dot(img, [65.481, 128.553, 24.966]) + 16.0
+    else:
+        out_img = np.matmul(
+            img, [[65.481, -37.797, 112.0], [128.553, -74.203, -93.786],
+                  [24.966, 112.0, -18.214]]) + [16, 128, 128]
+    out_img = _convert_output_type_range(out_img, img_type)
+    return out_img
+
+
+def bgr2ycbcr(img, y_only=False):
+    """Convert a BGR image to YCbCr image.
+
+    The bgr version of rgb2ycbcr.
+    It implements the ITU-R BT.601 conversion for standard-definition
+    television. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#ITU-R_BT.601_conversion.
+
+    It differs from a similar function in cv2.cvtColor: `BGR <-> YCrCb`.
+    In OpenCV, it implements a JPEG conversion. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#JPEG_conversion.
+
+    Args:
+        img (ndarray): The input image. It accepts:
+            1. np.uint8 type with range [0, 255];
+            2. np.float32 type with range [0, 1].
+        y_only (bool): Whether to only return Y channel. Default: False.
+
+    Returns:
+        ndarray: The converted YCbCr image. The output image has the same type
+            and range as input image.
+    """
+    img_type = img.dtype
+    img = _convert_input_type_range(img)
+    if y_only:
+        out_img = np.dot(img, [24.966, 128.553, 65.481]) + 16.0
+    else:
+        out_img = np.matmul(
+            img, [[24.966, 112.0, -18.214], [128.553, -74.203, -93.786],
+                  [65.481, -37.797, 112.0]]) + [16, 128, 128]
+    out_img = _convert_output_type_range(out_img, img_type)
+    return out_img
+
+
+def ycbcr2rgb(img):
+    """Convert a YCbCr image to RGB image.
+
+    This function produces the same results as Matlab's ycbcr2rgb function.
+    It implements the ITU-R BT.601 conversion for standard-definition
+    television. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#ITU-R_BT.601_conversion.
+
+    It differs from a similar function in cv2.cvtColor: `YCrCb <-> RGB`.
+    In OpenCV, it implements a JPEG conversion. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#JPEG_conversion.
+
+    Args:
+        img (ndarray): The input image. It accepts:
+            1. np.uint8 type with range [0, 255];
+            2. np.float32 type with range [0, 1].
+
+    Returns:
+        ndarray: The converted RGB image. The output image has the same type
+            and range as input image.
+    """
+    img_type = img.dtype
+    img = _convert_input_type_range(img) * 255
+    out_img = np.matmul(img, [[0.00456621, 0.00456621, 0.00456621],
+                              [0, -0.00153632, 0.00791071],
+                              [0.00625893, -0.00318811, 0]]) * 255.0 + [
+                                  -222.921, 135.576, -276.836
+                              ]
+    out_img = _convert_output_type_range(out_img, img_type)
+    return out_img
+
+
+def ycbcr2bgr(img):
+    """Convert a YCbCr image to BGR image.
+
+    The bgr version of ycbcr2rgb.
+    It implements the ITU-R BT.601 conversion for standard-definition
+    television. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#ITU-R_BT.601_conversion.
+
+    It differs from a similar function in cv2.cvtColor: `YCrCb <-> BGR`.
+    In OpenCV, it implements a JPEG conversion. See more details in
+    https://en.wikipedia.org/wiki/YCbCr#JPEG_conversion.
+
+    Args:
+        img (ndarray): The input image. It accepts:
+            1. np.uint8 type with range [0, 255];
+            2. np.float32 type with range [0, 1].
+
+    Returns:
+        ndarray: The converted BGR image. The output image has the same type
+            and range as input image.
+    """
+    img_type = img.dtype
+    img = _convert_input_type_range(img) * 255
+    out_img = np.matmul(img, [[0.00456621, 0.00456621, 0.00456621],
+                              [0.00791071, -0.00153632, 0],
+                              [0, -0.00318811, 0.00625893]]) * 255.0 + [
+                                  -276.836, 135.576, -222.921
+                              ]
+    out_img = _convert_output_type_range(out_img, img_type)
+    return out_img
+
+
+def convert_color_factory(src, dst):
+
+    code = getattr(cv2, f'COLOR_{src.upper()}2{dst.upper()}')
+
+    def convert_color(img):
+        out_img = cv2.cvtColor(img, code)
+        return out_img
+
+    convert_color.__doc__ = f"""Convert a {src.upper()} image to {dst.upper()}
+        image.
+
+    Args:
+        img (ndarray or str): The input image.
+
+    Returns:
+        ndarray: The converted {dst.upper()} image.
+    """
+
+    return convert_color
+
+
+bgr2rgb = convert_color_factory('bgr', 'rgb')
+
+rgb2bgr = convert_color_factory('rgb', 'bgr')
+
+bgr2hsv = convert_color_factory('bgr', 'hsv')
+
+hsv2bgr = convert_color_factory('hsv', 'bgr')
+
+bgr2hls = convert_color_factory('bgr', 'hls')
+
+hls2bgr = convert_color_factory('hls', 'bgr')
diff --git a/annotator/uniformer/mmcv/image/geometric.py b/annotator/uniformer/mmcv/image/geometric.py
new file mode 100644
index 0000000000000000000000000000000000000000..cf97c201cb4e43796c911919d03fb26a07ed817d
--- /dev/null
+++ b/annotator/uniformer/mmcv/image/geometric.py
@@ -0,0 +1,728 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import numbers
+
+import cv2
+import numpy as np
+
+from ..utils import to_2tuple
+from .io import imread_backend
+
+try:
+    from PIL import Image
+except ImportError:
+    Image = None
+
+
+def _scale_size(size, scale):
+    """Rescale a size by a ratio.
+
+    Args:
+        size (tuple[int]): (w, h).
+        scale (float | tuple(float)): Scaling factor.
+
+    Returns:
+        tuple[int]: scaled size.
+    """
+    if isinstance(scale, (float, int)):
+        scale = (scale, scale)
+    w, h = size
+    return int(w * float(scale[0]) + 0.5), int(h * float(scale[1]) + 0.5)
+
+
+cv2_interp_codes = {
+    'nearest': cv2.INTER_NEAREST,
+    'bilinear': cv2.INTER_LINEAR,
+    'bicubic': cv2.INTER_CUBIC,
+    'area': cv2.INTER_AREA,
+    'lanczos': cv2.INTER_LANCZOS4
+}
+
+if Image is not None:
+    pillow_interp_codes = {
+        'nearest': Image.NEAREST,
+        'bilinear': Image.BILINEAR,
+        'bicubic': Image.BICUBIC,
+        'box': Image.BOX,
+        'lanczos': Image.LANCZOS,
+        'hamming': Image.HAMMING
+    }
+
+
+def imresize(img,
+             size,
+             return_scale=False,
+             interpolation='bilinear',
+             out=None,
+             backend=None):
+    """Resize image to a given size.
+
+    Args:
+        img (ndarray): The input image.
+        size (tuple[int]): Target size (w, h).
+        return_scale (bool): Whether to return `w_scale` and `h_scale`.
+        interpolation (str): Interpolation method, accepted values are
+            "nearest", "bilinear", "bicubic", "area", "lanczos" for 'cv2'
+            backend, "nearest", "bilinear" for 'pillow' backend.
+        out (ndarray): The output destination.
+        backend (str | None): The image resize backend type. Options are `cv2`,
+            `pillow`, `None`. If backend is None, the global imread_backend
+            specified by ``mmcv.use_backend()`` will be used. Default: None.
+
+    Returns:
+        tuple | ndarray: (`resized_img`, `w_scale`, `h_scale`) or
+            `resized_img`.
+    """
+    h, w = img.shape[:2]
+    if backend is None:
+        backend = imread_backend
+    if backend not in ['cv2', 'pillow']:
+        raise ValueError(f'backend: {backend} is not supported for resize.'
+                         f"Supported backends are 'cv2', 'pillow'")
+
+    if backend == 'pillow':
+        assert img.dtype == np.uint8, 'Pillow backend only support uint8 type'
+        pil_image = Image.fromarray(img)
+        pil_image = pil_image.resize(size, pillow_interp_codes[interpolation])
+        resized_img = np.array(pil_image)
+    else:
+        resized_img = cv2.resize(
+            img, size, dst=out, interpolation=cv2_interp_codes[interpolation])
+    if not return_scale:
+        return resized_img
+    else:
+        w_scale = size[0] / w
+        h_scale = size[1] / h
+        return resized_img, w_scale, h_scale
+
+
+def imresize_to_multiple(img,
+                         divisor,
+                         size=None,
+                         scale_factor=None,
+                         keep_ratio=False,
+                         return_scale=False,
+                         interpolation='bilinear',
+                         out=None,
+                         backend=None):
+    """Resize image according to a given size or scale factor and then rounds
+    up the the resized or rescaled image size to the nearest value that can be
+    divided by the divisor.
+
+    Args:
+        img (ndarray): The input image.
+        divisor (int | tuple): Resized image size will be a multiple of
+            divisor. If divisor is a tuple, divisor should be
+            (w_divisor, h_divisor).
+        size (None | int | tuple[int]): Target size (w, h). Default: None.
+        scale_factor (None | float | tuple[float]): Multiplier for spatial
+            size. Should match input size if it is a tuple and the 2D style is
+            (w_scale_factor, h_scale_factor). Default: None.
+        keep_ratio (bool): Whether to keep the aspect ratio when resizing the
+            image. Default: False.
+        return_scale (bool): Whether to return `w_scale` and `h_scale`.
+        interpolation (str): Interpolation method, accepted values are
+            "nearest", "bilinear", "bicubic", "area", "lanczos" for 'cv2'
+            backend, "nearest", "bilinear" for 'pillow' backend.
+        out (ndarray): The output destination.
+        backend (str | None): The image resize backend type. Options are `cv2`,
+            `pillow`, `None`. If backend is None, the global imread_backend
+            specified by ``mmcv.use_backend()`` will be used. Default: None.
+
+    Returns:
+        tuple | ndarray: (`resized_img`, `w_scale`, `h_scale`) or
+            `resized_img`.
+    """
+    h, w = img.shape[:2]
+    if size is not None and scale_factor is not None:
+        raise ValueError('only one of size or scale_factor should be defined')
+    elif size is None and scale_factor is None:
+        raise ValueError('one of size or scale_factor should be defined')
+    elif size is not None:
+        size = to_2tuple(size)
+        if keep_ratio:
+            size = rescale_size((w, h), size, return_scale=False)
+    else:
+        size = _scale_size((w, h), scale_factor)
+
+    divisor = to_2tuple(divisor)
+    size = tuple([int(np.ceil(s / d)) * d for s, d in zip(size, divisor)])
+    resized_img, w_scale, h_scale = imresize(
+        img,
+        size,
+        return_scale=True,
+        interpolation=interpolation,
+        out=out,
+        backend=backend)
+    if return_scale:
+        return resized_img, w_scale, h_scale
+    else:
+        return resized_img
+
+
+def imresize_like(img,
+                  dst_img,
+                  return_scale=False,
+                  interpolation='bilinear',
+                  backend=None):
+    """Resize image to the same size of a given image.
+
+    Args:
+        img (ndarray): The input image.
+        dst_img (ndarray): The target image.
+        return_scale (bool): Whether to return `w_scale` and `h_scale`.
+        interpolation (str): Same as :func:`resize`.
+        backend (str | None): Same as :func:`resize`.
+
+    Returns:
+        tuple or ndarray: (`resized_img`, `w_scale`, `h_scale`) or
+            `resized_img`.
+    """
+    h, w = dst_img.shape[:2]
+    return imresize(img, (w, h), return_scale, interpolation, backend=backend)
+
+
+def rescale_size(old_size, scale, return_scale=False):
+    """Calculate the new size to be rescaled to.
+
+    Args:
+        old_size (tuple[int]): The old size (w, h) of image.
+        scale (float | tuple[int]): The scaling factor or maximum size.
+            If it is a float number, then the image will be rescaled by this
+            factor, else if it is a tuple of 2 integers, then the image will
+            be rescaled as large as possible within the scale.
+        return_scale (bool): Whether to return the scaling factor besides the
+            rescaled image size.
+
+    Returns:
+        tuple[int]: The new rescaled image size.
+    """
+    w, h = old_size
+    if isinstance(scale, (float, int)):
+        if scale <= 0:
+            raise ValueError(f'Invalid scale {scale}, must be positive.')
+        scale_factor = scale
+    elif isinstance(scale, tuple):
+        max_long_edge = max(scale)
+        max_short_edge = min(scale)
+        scale_factor = min(max_long_edge / max(h, w),
+                           max_short_edge / min(h, w))
+    else:
+        raise TypeError(
+            f'Scale must be a number or tuple of int, but got {type(scale)}')
+
+    new_size = _scale_size((w, h), scale_factor)
+
+    if return_scale:
+        return new_size, scale_factor
+    else:
+        return new_size
+
+
+def imrescale(img,
+              scale,
+              return_scale=False,
+              interpolation='bilinear',
+              backend=None):
+    """Resize image while keeping the aspect ratio.
+
+    Args:
+        img (ndarray): The input image.
+        scale (float | tuple[int]): The scaling factor or maximum size.
+            If it is a float number, then the image will be rescaled by this
+            factor, else if it is a tuple of 2 integers, then the image will
+            be rescaled as large as possible within the scale.
+        return_scale (bool): Whether to return the scaling factor besides the
+            rescaled image.
+        interpolation (str): Same as :func:`resize`.
+        backend (str | None): Same as :func:`resize`.
+
+    Returns:
+        ndarray: The rescaled image.
+    """
+    h, w = img.shape[:2]
+    new_size, scale_factor = rescale_size((w, h), scale, return_scale=True)
+    rescaled_img = imresize(
+        img, new_size, interpolation=interpolation, backend=backend)
+    if return_scale:
+        return rescaled_img, scale_factor
+    else:
+        return rescaled_img
+
+
+def imflip(img, direction='horizontal'):
+    """Flip an image horizontally or vertically.
+
+    Args:
+        img (ndarray): Image to be flipped.
+        direction (str): The flip direction, either "horizontal" or
+            "vertical" or "diagonal".
+
+    Returns:
+        ndarray: The flipped image.
+    """
+    assert direction in ['horizontal', 'vertical', 'diagonal']
+    if direction == 'horizontal':
+        return np.flip(img, axis=1)
+    elif direction == 'vertical':
+        return np.flip(img, axis=0)
+    else:
+        return np.flip(img, axis=(0, 1))
+
+
+def imflip_(img, direction='horizontal'):
+    """Inplace flip an image horizontally or vertically.
+
+    Args:
+        img (ndarray): Image to be flipped.
+        direction (str): The flip direction, either "horizontal" or
+            "vertical" or "diagonal".
+
+    Returns:
+        ndarray: The flipped image (inplace).
+    """
+    assert direction in ['horizontal', 'vertical', 'diagonal']
+    if direction == 'horizontal':
+        return cv2.flip(img, 1, img)
+    elif direction == 'vertical':
+        return cv2.flip(img, 0, img)
+    else:
+        return cv2.flip(img, -1, img)
+
+
+def imrotate(img,
+             angle,
+             center=None,
+             scale=1.0,
+             border_value=0,
+             interpolation='bilinear',
+             auto_bound=False):
+    """Rotate an image.
+
+    Args:
+        img (ndarray): Image to be rotated.
+        angle (float): Rotation angle in degrees, positive values mean
+            clockwise rotation.
+        center (tuple[float], optional): Center point (w, h) of the rotation in
+            the source image. If not specified, the center of the image will be
+            used.
+        scale (float): Isotropic scale factor.
+        border_value (int): Border value.
+        interpolation (str): Same as :func:`resize`.
+        auto_bound (bool): Whether to adjust the image size to cover the whole
+            rotated image.
+
+    Returns:
+        ndarray: The rotated image.
+    """
+    if center is not None and auto_bound:
+        raise ValueError('`auto_bound` conflicts with `center`')
+    h, w = img.shape[:2]
+    if center is None:
+        center = ((w - 1) * 0.5, (h - 1) * 0.5)
+    assert isinstance(center, tuple)
+
+    matrix = cv2.getRotationMatrix2D(center, -angle, scale)
+    if auto_bound:
+        cos = np.abs(matrix[0, 0])
+        sin = np.abs(matrix[0, 1])
+        new_w = h * sin + w * cos
+        new_h = h * cos + w * sin
+        matrix[0, 2] += (new_w - w) * 0.5
+        matrix[1, 2] += (new_h - h) * 0.5
+        w = int(np.round(new_w))
+        h = int(np.round(new_h))
+    rotated = cv2.warpAffine(
+        img,
+        matrix, (w, h),
+        flags=cv2_interp_codes[interpolation],
+        borderValue=border_value)
+    return rotated
+
+
+def bbox_clip(bboxes, img_shape):
+    """Clip bboxes to fit the image shape.
+
+    Args:
+        bboxes (ndarray): Shape (..., 4*k)
+        img_shape (tuple[int]): (height, width) of the image.
+
+    Returns:
+        ndarray: Clipped bboxes.
+    """
+    assert bboxes.shape[-1] % 4 == 0
+    cmin = np.empty(bboxes.shape[-1], dtype=bboxes.dtype)
+    cmin[0::2] = img_shape[1] - 1
+    cmin[1::2] = img_shape[0] - 1
+    clipped_bboxes = np.maximum(np.minimum(bboxes, cmin), 0)
+    return clipped_bboxes
+
+
+def bbox_scaling(bboxes, scale, clip_shape=None):
+    """Scaling bboxes w.r.t the box center.
+
+    Args:
+        bboxes (ndarray): Shape(..., 4).
+        scale (float): Scaling factor.
+        clip_shape (tuple[int], optional): If specified, bboxes that exceed the
+            boundary will be clipped according to the given shape (h, w).
+
+    Returns:
+        ndarray: Scaled bboxes.
+    """
+    if float(scale) == 1.0:
+        scaled_bboxes = bboxes.copy()
+    else:
+        w = bboxes[..., 2] - bboxes[..., 0] + 1
+        h = bboxes[..., 3] - bboxes[..., 1] + 1
+        dw = (w * (scale - 1)) * 0.5
+        dh = (h * (scale - 1)) * 0.5
+        scaled_bboxes = bboxes + np.stack((-dw, -dh, dw, dh), axis=-1)
+    if clip_shape is not None:
+        return bbox_clip(scaled_bboxes, clip_shape)
+    else:
+        return scaled_bboxes
+
+
+def imcrop(img, bboxes, scale=1.0, pad_fill=None):
+    """Crop image patches.
+
+    3 steps: scale the bboxes -> clip bboxes -> crop and pad.
+
+    Args:
+        img (ndarray): Image to be cropped.
+        bboxes (ndarray): Shape (k, 4) or (4, ), location of cropped bboxes.
+        scale (float, optional): Scale ratio of bboxes, the default value
+            1.0 means no padding.
+        pad_fill (Number | list[Number]): Value to be filled for padding.
+            Default: None, which means no padding.
+
+    Returns:
+        list[ndarray] | ndarray: The cropped image patches.
+    """
+    chn = 1 if img.ndim == 2 else img.shape[2]
+    if pad_fill is not None:
+        if isinstance(pad_fill, (int, float)):
+            pad_fill = [pad_fill for _ in range(chn)]
+        assert len(pad_fill) == chn
+
+    _bboxes = bboxes[None, ...] if bboxes.ndim == 1 else bboxes
+    scaled_bboxes = bbox_scaling(_bboxes, scale).astype(np.int32)
+    clipped_bbox = bbox_clip(scaled_bboxes, img.shape)
+
+    patches = []
+    for i in range(clipped_bbox.shape[0]):
+        x1, y1, x2, y2 = tuple(clipped_bbox[i, :])
+        if pad_fill is None:
+            patch = img[y1:y2 + 1, x1:x2 + 1, ...]
+        else:
+            _x1, _y1, _x2, _y2 = tuple(scaled_bboxes[i, :])
+            if chn == 1:
+                patch_shape = (_y2 - _y1 + 1, _x2 - _x1 + 1)
+            else:
+                patch_shape = (_y2 - _y1 + 1, _x2 - _x1 + 1, chn)
+            patch = np.array(
+                pad_fill, dtype=img.dtype) * np.ones(
+                    patch_shape, dtype=img.dtype)
+            x_start = 0 if _x1 >= 0 else -_x1
+            y_start = 0 if _y1 >= 0 else -_y1
+            w = x2 - x1 + 1
+            h = y2 - y1 + 1
+            patch[y_start:y_start + h, x_start:x_start + w,
+                  ...] = img[y1:y1 + h, x1:x1 + w, ...]
+        patches.append(patch)
+
+    if bboxes.ndim == 1:
+        return patches[0]
+    else:
+        return patches
+
+
+def impad(img,
+          *,
+          shape=None,
+          padding=None,
+          pad_val=0,
+          padding_mode='constant'):
+    """Pad the given image to a certain shape or pad on all sides with
+    specified padding mode and padding value.
+
+    Args:
+        img (ndarray): Image to be padded.
+        shape (tuple[int]): Expected padding shape (h, w). Default: None.
+        padding (int or tuple[int]): Padding on each border. If a single int is
+            provided this is used to pad all borders. If tuple of length 2 is
+            provided this is the padding on left/right and top/bottom
+            respectively. If a tuple of length 4 is provided this is the
+            padding for the left, top, right and bottom borders respectively.
+            Default: None. Note that `shape` and `padding` can not be both
+            set.
+        pad_val (Number | Sequence[Number]): Values to be filled in padding
+            areas when padding_mode is 'constant'. Default: 0.
+        padding_mode (str): Type of padding. Should be: constant, edge,
+            reflect or symmetric. Default: constant.
+
+            - constant: pads with a constant value, this value is specified
+                with pad_val.
+            - edge: pads with the last value at the edge of the image.
+            - reflect: pads with reflection of image without repeating the
+                last value on the edge. For example, padding [1, 2, 3, 4]
+                with 2 elements on both sides in reflect mode will result
+                in [3, 2, 1, 2, 3, 4, 3, 2].
+            - symmetric: pads with reflection of image repeating the last
+                value on the edge. For example, padding [1, 2, 3, 4] with
+                2 elements on both sides in symmetric mode will result in
+                [2, 1, 1, 2, 3, 4, 4, 3]
+
+    Returns:
+        ndarray: The padded image.
+    """
+
+    assert (shape is not None) ^ (padding is not None)
+    if shape is not None:
+        padding = (0, 0, shape[1] - img.shape[1], shape[0] - img.shape[0])
+
+    # check pad_val
+    if isinstance(pad_val, tuple):
+        assert len(pad_val) == img.shape[-1]
+    elif not isinstance(pad_val, numbers.Number):
+        raise TypeError('pad_val must be a int or a tuple. '
+                        f'But received {type(pad_val)}')
+
+    # check padding
+    if isinstance(padding, tuple) and len(padding) in [2, 4]:
+        if len(padding) == 2:
+            padding = (padding[0], padding[1], padding[0], padding[1])
+    elif isinstance(padding, numbers.Number):
+        padding = (padding, padding, padding, padding)
+    else:
+        raise ValueError('Padding must be a int or a 2, or 4 element tuple.'
+                         f'But received {padding}')
+
+    # check padding mode
+    assert padding_mode in ['constant', 'edge', 'reflect', 'symmetric']
+
+    border_type = {
+        'constant': cv2.BORDER_CONSTANT,
+        'edge': cv2.BORDER_REPLICATE,
+        'reflect': cv2.BORDER_REFLECT_101,
+        'symmetric': cv2.BORDER_REFLECT
+    }
+    img = cv2.copyMakeBorder(
+        img,
+        padding[1],
+        padding[3],
+        padding[0],
+        padding[2],
+        border_type[padding_mode],
+        value=pad_val)
+
+    return img
+
+
+def impad_to_multiple(img, divisor, pad_val=0):
+    """Pad an image to ensure each edge to be multiple to some number.
+
+    Args:
+        img (ndarray): Image to be padded.
+        divisor (int): Padded image edges will be multiple to divisor.
+        pad_val (Number | Sequence[Number]): Same as :func:`impad`.
+
+    Returns:
+        ndarray: The padded image.
+    """
+    pad_h = int(np.ceil(img.shape[0] / divisor)) * divisor
+    pad_w = int(np.ceil(img.shape[1] / divisor)) * divisor
+    return impad(img, shape=(pad_h, pad_w), pad_val=pad_val)
+
+
+def cutout(img, shape, pad_val=0):
+    """Randomly cut out a rectangle from the original img.
+
+    Args:
+        img (ndarray): Image to be cutout.
+        shape (int | tuple[int]): Expected cutout shape (h, w). If given as a
+            int, the value will be used for both h and w.
+        pad_val (int | float | tuple[int | float]): Values to be filled in the
+            cut area. Defaults to 0.
+
+    Returns:
+        ndarray: The cutout image.
+    """
+
+    channels = 1 if img.ndim == 2 else img.shape[2]
+    if isinstance(shape, int):
+        cut_h, cut_w = shape, shape
+    else:
+        assert isinstance(shape, tuple) and len(shape) == 2, \
+            f'shape must be a int or a tuple with length 2, but got type ' \
+            f'{type(shape)} instead.'
+        cut_h, cut_w = shape
+    if isinstance(pad_val, (int, float)):
+        pad_val = tuple([pad_val] * channels)
+    elif isinstance(pad_val, tuple):
+        assert len(pad_val) == channels, \
+            'Expected the num of elements in tuple equals the channels' \
+            'of input image. Found {} vs {}'.format(
+                len(pad_val), channels)
+    else:
+        raise TypeError(f'Invalid type {type(pad_val)} for `pad_val`')
+
+    img_h, img_w = img.shape[:2]
+    y0 = np.random.uniform(img_h)
+    x0 = np.random.uniform(img_w)
+
+    y1 = int(max(0, y0 - cut_h / 2.))
+    x1 = int(max(0, x0 - cut_w / 2.))
+    y2 = min(img_h, y1 + cut_h)
+    x2 = min(img_w, x1 + cut_w)
+
+    if img.ndim == 2:
+        patch_shape = (y2 - y1, x2 - x1)
+    else:
+        patch_shape = (y2 - y1, x2 - x1, channels)
+
+    img_cutout = img.copy()
+    patch = np.array(
+        pad_val, dtype=img.dtype) * np.ones(
+            patch_shape, dtype=img.dtype)
+    img_cutout[y1:y2, x1:x2, ...] = patch
+
+    return img_cutout
+
+
+def _get_shear_matrix(magnitude, direction='horizontal'):
+    """Generate the shear matrix for transformation.
+
+    Args:
+        magnitude (int | float): The magnitude used for shear.
+        direction (str): The flip direction, either "horizontal"
+            or "vertical".
+
+    Returns:
+        ndarray: The shear matrix with dtype float32.
+    """
+    if direction == 'horizontal':
+        shear_matrix = np.float32([[1, magnitude, 0], [0, 1, 0]])
+    elif direction == 'vertical':
+        shear_matrix = np.float32([[1, 0, 0], [magnitude, 1, 0]])
+    return shear_matrix
+
+
+def imshear(img,
+            magnitude,
+            direction='horizontal',
+            border_value=0,
+            interpolation='bilinear'):
+    """Shear an image.
+
+    Args:
+        img (ndarray): Image to be sheared with format (h, w)
+            or (h, w, c).
+        magnitude (int | float): The magnitude used for shear.
+        direction (str): The flip direction, either "horizontal"
+            or "vertical".
+        border_value (int | tuple[int]): Value used in case of a
+            constant border.
+        interpolation (str): Same as :func:`resize`.
+
+    Returns:
+        ndarray: The sheared image.
+    """
+    assert direction in ['horizontal',
+                         'vertical'], f'Invalid direction: {direction}'
+    height, width = img.shape[:2]
+    if img.ndim == 2:
+        channels = 1
+    elif img.ndim == 3:
+        channels = img.shape[-1]
+    if isinstance(border_value, int):
+        border_value = tuple([border_value] * channels)
+    elif isinstance(border_value, tuple):
+        assert len(border_value) == channels, \
+            'Expected the num of elements in tuple equals the channels' \
+            'of input image. Found {} vs {}'.format(
+                len(border_value), channels)
+    else:
+        raise ValueError(
+            f'Invalid type {type(border_value)} for `border_value`')
+    shear_matrix = _get_shear_matrix(magnitude, direction)
+    sheared = cv2.warpAffine(
+        img,
+        shear_matrix,
+        (width, height),
+        # Note case when the number elements in `border_value`
+        # greater than 3 (e.g. shearing masks whose channels large
+        # than 3) will raise TypeError in `cv2.warpAffine`.
+        # Here simply slice the first 3 values in `border_value`.
+        borderValue=border_value[:3],
+        flags=cv2_interp_codes[interpolation])
+    return sheared
+
+
+def _get_translate_matrix(offset, direction='horizontal'):
+    """Generate the translate matrix.
+
+    Args:
+        offset (int | float): The offset used for translate.
+        direction (str): The translate direction, either
+            "horizontal" or "vertical".
+
+    Returns:
+        ndarray: The translate matrix with dtype float32.
+    """
+    if direction == 'horizontal':
+        translate_matrix = np.float32([[1, 0, offset], [0, 1, 0]])
+    elif direction == 'vertical':
+        translate_matrix = np.float32([[1, 0, 0], [0, 1, offset]])
+    return translate_matrix
+
+
+def imtranslate(img,
+                offset,
+                direction='horizontal',
+                border_value=0,
+                interpolation='bilinear'):
+    """Translate an image.
+
+    Args:
+        img (ndarray): Image to be translated with format
+            (h, w) or (h, w, c).
+        offset (int | float): The offset used for translate.
+        direction (str): The translate direction, either "horizontal"
+            or "vertical".
+        border_value (int | tuple[int]): Value used in case of a
+            constant border.
+        interpolation (str): Same as :func:`resize`.
+
+    Returns:
+        ndarray: The translated image.
+    """
+    assert direction in ['horizontal',
+                         'vertical'], f'Invalid direction: {direction}'
+    height, width = img.shape[:2]
+    if img.ndim == 2:
+        channels = 1
+    elif img.ndim == 3:
+        channels = img.shape[-1]
+    if isinstance(border_value, int):
+        border_value = tuple([border_value] * channels)
+    elif isinstance(border_value, tuple):
+        assert len(border_value) == channels, \
+            'Expected the num of elements in tuple equals the channels' \
+            'of input image. Found {} vs {}'.format(
+                len(border_value), channels)
+    else:
+        raise ValueError(
+            f'Invalid type {type(border_value)} for `border_value`.')
+    translate_matrix = _get_translate_matrix(offset, direction)
+    translated = cv2.warpAffine(
+        img,
+        translate_matrix,
+        (width, height),
+        # Note case when the number elements in `border_value`
+        # greater than 3 (e.g. translating masks whose channels
+        # large than 3) will raise TypeError in `cv2.warpAffine`.
+        # Here simply slice the first 3 values in `border_value`.
+        borderValue=border_value[:3],
+        flags=cv2_interp_codes[interpolation])
+    return translated
diff --git a/annotator/uniformer/mmcv/image/io.py b/annotator/uniformer/mmcv/image/io.py
new file mode 100644
index 0000000000000000000000000000000000000000..d3fa2e8cc06b1a7b0b69de6406980b15d61a1e5d
--- /dev/null
+++ b/annotator/uniformer/mmcv/image/io.py
@@ -0,0 +1,258 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import io
+import os.path as osp
+from pathlib import Path
+
+import cv2
+import numpy as np
+from cv2 import (IMREAD_COLOR, IMREAD_GRAYSCALE, IMREAD_IGNORE_ORIENTATION,
+                 IMREAD_UNCHANGED)
+
+from annotator.uniformer.mmcv.utils import check_file_exist, is_str, mkdir_or_exist
+
+try:
+    from turbojpeg import TJCS_RGB, TJPF_BGR, TJPF_GRAY, TurboJPEG
+except ImportError:
+    TJCS_RGB = TJPF_GRAY = TJPF_BGR = TurboJPEG = None
+
+try:
+    from PIL import Image, ImageOps
+except ImportError:
+    Image = None
+
+try:
+    import tifffile
+except ImportError:
+    tifffile = None
+
+jpeg = None
+supported_backends = ['cv2', 'turbojpeg', 'pillow', 'tifffile']
+
+imread_flags = {
+    'color': IMREAD_COLOR,
+    'grayscale': IMREAD_GRAYSCALE,
+    'unchanged': IMREAD_UNCHANGED,
+    'color_ignore_orientation': IMREAD_IGNORE_ORIENTATION | IMREAD_COLOR,
+    'grayscale_ignore_orientation':
+    IMREAD_IGNORE_ORIENTATION | IMREAD_GRAYSCALE
+}
+
+imread_backend = 'cv2'
+
+
+def use_backend(backend):
+    """Select a backend for image decoding.
+
+    Args:
+        backend (str): The image decoding backend type. Options are `cv2`,
+        `pillow`, `turbojpeg` (see https://github.com/lilohuang/PyTurboJPEG)
+        and `tifffile`. `turbojpeg` is faster but it only supports `.jpeg`
+        file format.
+    """
+    assert backend in supported_backends
+    global imread_backend
+    imread_backend = backend
+    if imread_backend == 'turbojpeg':
+        if TurboJPEG is None:
+            raise ImportError('`PyTurboJPEG` is not installed')
+        global jpeg
+        if jpeg is None:
+            jpeg = TurboJPEG()
+    elif imread_backend == 'pillow':
+        if Image is None:
+            raise ImportError('`Pillow` is not installed')
+    elif imread_backend == 'tifffile':
+        if tifffile is None:
+            raise ImportError('`tifffile` is not installed')
+
+
+def _jpegflag(flag='color', channel_order='bgr'):
+    channel_order = channel_order.lower()
+    if channel_order not in ['rgb', 'bgr']:
+        raise ValueError('channel order must be either "rgb" or "bgr"')
+
+    if flag == 'color':
+        if channel_order == 'bgr':
+            return TJPF_BGR
+        elif channel_order == 'rgb':
+            return TJCS_RGB
+    elif flag == 'grayscale':
+        return TJPF_GRAY
+    else:
+        raise ValueError('flag must be "color" or "grayscale"')
+
+
+def _pillow2array(img, flag='color', channel_order='bgr'):
+    """Convert a pillow image to numpy array.
+
+    Args:
+        img (:obj:`PIL.Image.Image`): The image loaded using PIL
+        flag (str): Flags specifying the color type of a loaded image,
+            candidates are 'color', 'grayscale' and 'unchanged'.
+            Default to 'color'.
+        channel_order (str): The channel order of the output image array,
+            candidates are 'bgr' and 'rgb'. Default to 'bgr'.
+
+    Returns:
+        np.ndarray: The converted numpy array
+    """
+    channel_order = channel_order.lower()
+    if channel_order not in ['rgb', 'bgr']:
+        raise ValueError('channel order must be either "rgb" or "bgr"')
+
+    if flag == 'unchanged':
+        array = np.array(img)
+        if array.ndim >= 3 and array.shape[2] >= 3:  # color image
+            array[:, :, :3] = array[:, :, (2, 1, 0)]  # RGB to BGR
+    else:
+        # Handle exif orientation tag
+        if flag in ['color', 'grayscale']:
+            img = ImageOps.exif_transpose(img)
+        # If the image mode is not 'RGB', convert it to 'RGB' first.
+        if img.mode != 'RGB':
+            if img.mode != 'LA':
+                # Most formats except 'LA' can be directly converted to RGB
+                img = img.convert('RGB')
+            else:
+                # When the mode is 'LA', the default conversion will fill in
+                #  the canvas with black, which sometimes shadows black objects
+                #  in the foreground.
+                #
+                # Therefore, a random color (124, 117, 104) is used for canvas
+                img_rgba = img.convert('RGBA')
+                img = Image.new('RGB', img_rgba.size, (124, 117, 104))
+                img.paste(img_rgba, mask=img_rgba.split()[3])  # 3 is alpha
+        if flag in ['color', 'color_ignore_orientation']:
+            array = np.array(img)
+            if channel_order != 'rgb':
+                array = array[:, :, ::-1]  # RGB to BGR
+        elif flag in ['grayscale', 'grayscale_ignore_orientation']:
+            img = img.convert('L')
+            array = np.array(img)
+        else:
+            raise ValueError(
+                'flag must be "color", "grayscale", "unchanged", '
+                f'"color_ignore_orientation" or "grayscale_ignore_orientation"'
+                f' but got {flag}')
+    return array
+
+
+def imread(img_or_path, flag='color', channel_order='bgr', backend=None):
+    """Read an image.
+
+    Args:
+        img_or_path (ndarray or str or Path): Either a numpy array or str or
+            pathlib.Path. If it is a numpy array (loaded image), then
+            it will be returned as is.
+        flag (str): Flags specifying the color type of a loaded image,
+            candidates are `color`, `grayscale`, `unchanged`,
+            `color_ignore_orientation` and `grayscale_ignore_orientation`.
+            By default, `cv2` and `pillow` backend would rotate the image
+            according to its EXIF info unless called with `unchanged` or
+            `*_ignore_orientation` flags. `turbojpeg` and `tifffile` backend
+            always ignore image's EXIF info regardless of the flag.
+            The `turbojpeg` backend only supports `color` and `grayscale`.
+        channel_order (str): Order of channel, candidates are `bgr` and `rgb`.
+        backend (str | None): The image decoding backend type. Options are
+            `cv2`, `pillow`, `turbojpeg`, `tifffile`, `None`.
+            If backend is None, the global imread_backend specified by
+            ``mmcv.use_backend()`` will be used. Default: None.
+
+    Returns:
+        ndarray: Loaded image array.
+    """
+
+    if backend is None:
+        backend = imread_backend
+    if backend not in supported_backends:
+        raise ValueError(f'backend: {backend} is not supported. Supported '
+                         "backends are 'cv2', 'turbojpeg', 'pillow'")
+    if isinstance(img_or_path, Path):
+        img_or_path = str(img_or_path)
+
+    if isinstance(img_or_path, np.ndarray):
+        return img_or_path
+    elif is_str(img_or_path):
+        check_file_exist(img_or_path,
+                         f'img file does not exist: {img_or_path}')
+        if backend == 'turbojpeg':
+            with open(img_or_path, 'rb') as in_file:
+                img = jpeg.decode(in_file.read(),
+                                  _jpegflag(flag, channel_order))
+                if img.shape[-1] == 1:
+                    img = img[:, :, 0]
+            return img
+        elif backend == 'pillow':
+            img = Image.open(img_or_path)
+            img = _pillow2array(img, flag, channel_order)
+            return img
+        elif backend == 'tifffile':
+            img = tifffile.imread(img_or_path)
+            return img
+        else:
+            flag = imread_flags[flag] if is_str(flag) else flag
+            img = cv2.imread(img_or_path, flag)
+            if flag == IMREAD_COLOR and channel_order == 'rgb':
+                cv2.cvtColor(img, cv2.COLOR_BGR2RGB, img)
+            return img
+    else:
+        raise TypeError('"img" must be a numpy array or a str or '
+                        'a pathlib.Path object')
+
+
+def imfrombytes(content, flag='color', channel_order='bgr', backend=None):
+    """Read an image from bytes.
+
+    Args:
+        content (bytes): Image bytes got from files or other streams.
+        flag (str): Same as :func:`imread`.
+        backend (str | None): The image decoding backend type. Options are
+            `cv2`, `pillow`, `turbojpeg`, `None`. If backend is None, the
+            global imread_backend specified by ``mmcv.use_backend()`` will be
+            used. Default: None.
+
+    Returns:
+        ndarray: Loaded image array.
+    """
+
+    if backend is None:
+        backend = imread_backend
+    if backend not in supported_backends:
+        raise ValueError(f'backend: {backend} is not supported. Supported '
+                         "backends are 'cv2', 'turbojpeg', 'pillow'")
+    if backend == 'turbojpeg':
+        img = jpeg.decode(content, _jpegflag(flag, channel_order))
+        if img.shape[-1] == 1:
+            img = img[:, :, 0]
+        return img
+    elif backend == 'pillow':
+        buff = io.BytesIO(content)
+        img = Image.open(buff)
+        img = _pillow2array(img, flag, channel_order)
+        return img
+    else:
+        img_np = np.frombuffer(content, np.uint8)
+        flag = imread_flags[flag] if is_str(flag) else flag
+        img = cv2.imdecode(img_np, flag)
+        if flag == IMREAD_COLOR and channel_order == 'rgb':
+            cv2.cvtColor(img, cv2.COLOR_BGR2RGB, img)
+        return img
+
+
+def imwrite(img, file_path, params=None, auto_mkdir=True):
+    """Write image to file.
+
+    Args:
+        img (ndarray): Image array to be written.
+        file_path (str): Image file path.
+        params (None or list): Same as opencv :func:`imwrite` interface.
+        auto_mkdir (bool): If the parent folder of `file_path` does not exist,
+            whether to create it automatically.
+
+    Returns:
+        bool: Successful or not.
+    """
+    if auto_mkdir:
+        dir_name = osp.abspath(osp.dirname(file_path))
+        mkdir_or_exist(dir_name)
+    return cv2.imwrite(file_path, img, params)
diff --git a/annotator/uniformer/mmcv/image/misc.py b/annotator/uniformer/mmcv/image/misc.py
new file mode 100644
index 0000000000000000000000000000000000000000..3e61f05e3b05e4c7b40de4eb6c8eb100e6da41d0
--- /dev/null
+++ b/annotator/uniformer/mmcv/image/misc.py
@@ -0,0 +1,44 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import numpy as np
+
+import annotator.uniformer.mmcv as mmcv
+
+try:
+    import torch
+except ImportError:
+    torch = None
+
+
+def tensor2imgs(tensor, mean=(0, 0, 0), std=(1, 1, 1), to_rgb=True):
+    """Convert tensor to 3-channel images.
+
+    Args:
+        tensor (torch.Tensor): Tensor that contains multiple images, shape (
+            N, C, H, W).
+        mean (tuple[float], optional): Mean of images. Defaults to (0, 0, 0).
+        std (tuple[float], optional): Standard deviation of images.
+            Defaults to (1, 1, 1).
+        to_rgb (bool, optional): Whether the tensor was converted to RGB
+            format in the first place. If so, convert it back to BGR.
+            Defaults to True.
+
+    Returns:
+        list[np.ndarray]: A list that contains multiple images.
+    """
+
+    if torch is None:
+        raise RuntimeError('pytorch is not installed')
+    assert torch.is_tensor(tensor) and tensor.ndim == 4
+    assert len(mean) == 3
+    assert len(std) == 3
+
+    num_imgs = tensor.size(0)
+    mean = np.array(mean, dtype=np.float32)
+    std = np.array(std, dtype=np.float32)
+    imgs = []
+    for img_id in range(num_imgs):
+        img = tensor[img_id, ...].cpu().numpy().transpose(1, 2, 0)
+        img = mmcv.imdenormalize(
+            img, mean, std, to_bgr=to_rgb).astype(np.uint8)
+        imgs.append(np.ascontiguousarray(img))
+    return imgs
diff --git a/annotator/uniformer/mmcv/image/photometric.py b/annotator/uniformer/mmcv/image/photometric.py
new file mode 100644
index 0000000000000000000000000000000000000000..5085d012019c0cbf56f66f421a378278c1a058ae
--- /dev/null
+++ b/annotator/uniformer/mmcv/image/photometric.py
@@ -0,0 +1,428 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import cv2
+import numpy as np
+
+from ..utils import is_tuple_of
+from .colorspace import bgr2gray, gray2bgr
+
+
+def imnormalize(img, mean, std, to_rgb=True):
+    """Normalize an image with mean and std.
+
+    Args:
+        img (ndarray): Image to be normalized.
+        mean (ndarray): The mean to be used for normalize.
+        std (ndarray): The std to be used for normalize.
+        to_rgb (bool): Whether to convert to rgb.
+
+    Returns:
+        ndarray: The normalized image.
+    """
+    img = img.copy().astype(np.float32)
+    return imnormalize_(img, mean, std, to_rgb)
+
+
+def imnormalize_(img, mean, std, to_rgb=True):
+    """Inplace normalize an image with mean and std.
+
+    Args:
+        img (ndarray): Image to be normalized.
+        mean (ndarray): The mean to be used for normalize.
+        std (ndarray): The std to be used for normalize.
+        to_rgb (bool): Whether to convert to rgb.
+
+    Returns:
+        ndarray: The normalized image.
+    """
+    # cv2 inplace normalization does not accept uint8
+    assert img.dtype != np.uint8
+    mean = np.float64(mean.reshape(1, -1))
+    stdinv = 1 / np.float64(std.reshape(1, -1))
+    if to_rgb:
+        cv2.cvtColor(img, cv2.COLOR_BGR2RGB, img)  # inplace
+    cv2.subtract(img, mean, img)  # inplace
+    cv2.multiply(img, stdinv, img)  # inplace
+    return img
+
+
+def imdenormalize(img, mean, std, to_bgr=True):
+    assert img.dtype != np.uint8
+    mean = mean.reshape(1, -1).astype(np.float64)
+    std = std.reshape(1, -1).astype(np.float64)
+    img = cv2.multiply(img, std)  # make a copy
+    cv2.add(img, mean, img)  # inplace
+    if to_bgr:
+        cv2.cvtColor(img, cv2.COLOR_RGB2BGR, img)  # inplace
+    return img
+
+
+def iminvert(img):
+    """Invert (negate) an image.
+
+    Args:
+        img (ndarray): Image to be inverted.
+
+    Returns:
+        ndarray: The inverted image.
+    """
+    return np.full_like(img, 255) - img
+
+
+def solarize(img, thr=128):
+    """Solarize an image (invert all pixel values above a threshold)
+
+    Args:
+        img (ndarray): Image to be solarized.
+        thr (int): Threshold for solarizing (0 - 255).
+
+    Returns:
+        ndarray: The solarized image.
+    """
+    img = np.where(img < thr, img, 255 - img)
+    return img
+
+
+def posterize(img, bits):
+    """Posterize an image (reduce the number of bits for each color channel)
+
+    Args:
+        img (ndarray): Image to be posterized.
+        bits (int): Number of bits (1 to 8) to use for posterizing.
+
+    Returns:
+        ndarray: The posterized image.
+    """
+    shift = 8 - bits
+    img = np.left_shift(np.right_shift(img, shift), shift)
+    return img
+
+
+def adjust_color(img, alpha=1, beta=None, gamma=0):
+    r"""It blends the source image and its gray image:
+
+    .. math::
+        output = img * alpha + gray\_img * beta + gamma
+
+    Args:
+        img (ndarray): The input source image.
+        alpha (int | float): Weight for the source image. Default 1.
+        beta (int | float): Weight for the converted gray image.
+            If None, it's assigned the value (1 - `alpha`).
+        gamma (int | float): Scalar added to each sum.
+            Same as :func:`cv2.addWeighted`. Default 0.
+
+    Returns:
+        ndarray: Colored image which has the same size and dtype as input.
+    """
+    gray_img = bgr2gray(img)
+    gray_img = np.tile(gray_img[..., None], [1, 1, 3])
+    if beta is None:
+        beta = 1 - alpha
+    colored_img = cv2.addWeighted(img, alpha, gray_img, beta, gamma)
+    if not colored_img.dtype == np.uint8:
+        # Note when the dtype of `img` is not the default `np.uint8`
+        # (e.g. np.float32), the value in `colored_img` got from cv2
+        # is not guaranteed to be in range [0, 255], so here clip
+        # is needed.
+        colored_img = np.clip(colored_img, 0, 255)
+    return colored_img
+
+
+def imequalize(img):
+    """Equalize the image histogram.
+
+    This function applies a non-linear mapping to the input image,
+    in order to create a uniform distribution of grayscale values
+    in the output image.
+
+    Args:
+        img (ndarray): Image to be equalized.
+
+    Returns:
+        ndarray: The equalized image.
+    """
+
+    def _scale_channel(im, c):
+        """Scale the data in the corresponding channel."""
+        im = im[:, :, c]
+        # Compute the histogram of the image channel.
+        histo = np.histogram(im, 256, (0, 255))[0]
+        # For computing the step, filter out the nonzeros.
+        nonzero_histo = histo[histo > 0]
+        step = (np.sum(nonzero_histo) - nonzero_histo[-1]) // 255
+        if not step:
+            lut = np.array(range(256))
+        else:
+            # Compute the cumulative sum, shifted by step // 2
+            # and then normalized by step.
+            lut = (np.cumsum(histo) + (step // 2)) // step
+            # Shift lut, prepending with 0.
+            lut = np.concatenate([[0], lut[:-1]], 0)
+            # handle potential integer overflow
+            lut[lut > 255] = 255
+        # If step is zero, return the original image.
+        # Otherwise, index from lut.
+        return np.where(np.equal(step, 0), im, lut[im])
+
+    # Scales each channel independently and then stacks
+    # the result.
+    s1 = _scale_channel(img, 0)
+    s2 = _scale_channel(img, 1)
+    s3 = _scale_channel(img, 2)
+    equalized_img = np.stack([s1, s2, s3], axis=-1)
+    return equalized_img.astype(img.dtype)
+
+
+def adjust_brightness(img, factor=1.):
+    """Adjust image brightness.
+
+    This function controls the brightness of an image. An
+    enhancement factor of 0.0 gives a black image.
+    A factor of 1.0 gives the original image. This function
+    blends the source image and the degenerated black image:
+
+    .. math::
+        output = img * factor + degenerated * (1 - factor)
+
+    Args:
+        img (ndarray): Image to be brightened.
+        factor (float): A value controls the enhancement.
+            Factor 1.0 returns the original image, lower
+            factors mean less color (brightness, contrast,
+            etc), and higher values more. Default 1.
+
+    Returns:
+        ndarray: The brightened image.
+    """
+    degenerated = np.zeros_like(img)
+    # Note manually convert the dtype to np.float32, to
+    # achieve as close results as PIL.ImageEnhance.Brightness.
+    # Set beta=1-factor, and gamma=0
+    brightened_img = cv2.addWeighted(
+        img.astype(np.float32), factor, degenerated.astype(np.float32),
+        1 - factor, 0)
+    brightened_img = np.clip(brightened_img, 0, 255)
+    return brightened_img.astype(img.dtype)
+
+
+def adjust_contrast(img, factor=1.):
+    """Adjust image contrast.
+
+    This function controls the contrast of an image. An
+    enhancement factor of 0.0 gives a solid grey
+    image. A factor of 1.0 gives the original image. It
+    blends the source image and the degenerated mean image:
+
+    .. math::
+        output = img * factor + degenerated * (1 - factor)
+
+    Args:
+        img (ndarray): Image to be contrasted. BGR order.
+        factor (float): Same as :func:`mmcv.adjust_brightness`.
+
+    Returns:
+        ndarray: The contrasted image.
+    """
+    gray_img = bgr2gray(img)
+    hist = np.histogram(gray_img, 256, (0, 255))[0]
+    mean = round(np.sum(gray_img) / np.sum(hist))
+    degenerated = (np.ones_like(img[..., 0]) * mean).astype(img.dtype)
+    degenerated = gray2bgr(degenerated)
+    contrasted_img = cv2.addWeighted(
+        img.astype(np.float32), factor, degenerated.astype(np.float32),
+        1 - factor, 0)
+    contrasted_img = np.clip(contrasted_img, 0, 255)
+    return contrasted_img.astype(img.dtype)
+
+
+def auto_contrast(img, cutoff=0):
+    """Auto adjust image contrast.
+
+    This function maximize (normalize) image contrast by first removing cutoff
+    percent of the lightest and darkest pixels from the histogram and remapping
+    the image so that the darkest pixel becomes black (0), and the lightest
+    becomes white (255).
+
+    Args:
+        img (ndarray): Image to be contrasted. BGR order.
+        cutoff (int | float | tuple): The cutoff percent of the lightest and
+            darkest pixels to be removed. If given as tuple, it shall be
+            (low, high). Otherwise, the single value will be used for both.
+            Defaults to 0.
+
+    Returns:
+        ndarray: The contrasted image.
+    """
+
+    def _auto_contrast_channel(im, c, cutoff):
+        im = im[:, :, c]
+        # Compute the histogram of the image channel.
+        histo = np.histogram(im, 256, (0, 255))[0]
+        # Remove cut-off percent pixels from histo
+        histo_sum = np.cumsum(histo)
+        cut_low = histo_sum[-1] * cutoff[0] // 100
+        cut_high = histo_sum[-1] - histo_sum[-1] * cutoff[1] // 100
+        histo_sum = np.clip(histo_sum, cut_low, cut_high) - cut_low
+        histo = np.concatenate([[histo_sum[0]], np.diff(histo_sum)], 0)
+
+        # Compute mapping
+        low, high = np.nonzero(histo)[0][0], np.nonzero(histo)[0][-1]
+        # If all the values have been cut off, return the origin img
+        if low >= high:
+            return im
+        scale = 255.0 / (high - low)
+        offset = -low * scale
+        lut = np.array(range(256))
+        lut = lut * scale + offset
+        lut = np.clip(lut, 0, 255)
+        return lut[im]
+
+    if isinstance(cutoff, (int, float)):
+        cutoff = (cutoff, cutoff)
+    else:
+        assert isinstance(cutoff, tuple), 'cutoff must be of type int, ' \
+            f'float or tuple, but got {type(cutoff)} instead.'
+    # Auto adjusts contrast for each channel independently and then stacks
+    # the result.
+    s1 = _auto_contrast_channel(img, 0, cutoff)
+    s2 = _auto_contrast_channel(img, 1, cutoff)
+    s3 = _auto_contrast_channel(img, 2, cutoff)
+    contrasted_img = np.stack([s1, s2, s3], axis=-1)
+    return contrasted_img.astype(img.dtype)
+
+
+def adjust_sharpness(img, factor=1., kernel=None):
+    """Adjust image sharpness.
+
+    This function controls the sharpness of an image. An
+    enhancement factor of 0.0 gives a blurred image. A
+    factor of 1.0 gives the original image. And a factor
+    of 2.0 gives a sharpened image. It blends the source
+    image and the degenerated mean image:
+
+    .. math::
+        output = img * factor + degenerated * (1 - factor)
+
+    Args:
+        img (ndarray): Image to be sharpened. BGR order.
+        factor (float): Same as :func:`mmcv.adjust_brightness`.
+        kernel (np.ndarray, optional): Filter kernel to be applied on the img
+            to obtain the degenerated img. Defaults to None.
+
+    Note:
+        No value sanity check is enforced on the kernel set by users. So with
+        an inappropriate kernel, the ``adjust_sharpness`` may fail to perform
+        the function its name indicates but end up performing whatever
+        transform determined by the kernel.
+
+    Returns:
+        ndarray: The sharpened image.
+    """
+
+    if kernel is None:
+        # adopted from PIL.ImageFilter.SMOOTH
+        kernel = np.array([[1., 1., 1.], [1., 5., 1.], [1., 1., 1.]]) / 13
+    assert isinstance(kernel, np.ndarray), \
+        f'kernel must be of type np.ndarray, but got {type(kernel)} instead.'
+    assert kernel.ndim == 2, \
+        f'kernel must have a dimension of 2, but got {kernel.ndim} instead.'
+
+    degenerated = cv2.filter2D(img, -1, kernel)
+    sharpened_img = cv2.addWeighted(
+        img.astype(np.float32), factor, degenerated.astype(np.float32),
+        1 - factor, 0)
+    sharpened_img = np.clip(sharpened_img, 0, 255)
+    return sharpened_img.astype(img.dtype)
+
+
+def adjust_lighting(img, eigval, eigvec, alphastd=0.1, to_rgb=True):
+    """AlexNet-style PCA jitter.
+
+    This data augmentation is proposed in `ImageNet Classification with Deep
+    Convolutional Neural Networks
+    <https://dl.acm.org/doi/pdf/10.1145/3065386>`_.
+
+    Args:
+        img (ndarray): Image to be adjusted lighting. BGR order.
+        eigval (ndarray): the eigenvalue of the convariance matrix of pixel
+            values, respectively.
+        eigvec (ndarray): the eigenvector of the convariance matrix of pixel
+            values, respectively.
+        alphastd (float): The standard deviation for distribution of alpha.
+            Defaults to 0.1
+        to_rgb (bool): Whether to convert img to rgb.
+
+    Returns:
+        ndarray: The adjusted image.
+    """
+    assert isinstance(eigval, np.ndarray) and isinstance(eigvec, np.ndarray), \
+        f'eigval and eigvec should both be of type np.ndarray, got ' \
+        f'{type(eigval)} and {type(eigvec)} instead.'
+
+    assert eigval.ndim == 1 and eigvec.ndim == 2
+    assert eigvec.shape == (3, eigval.shape[0])
+    n_eigval = eigval.shape[0]
+    assert isinstance(alphastd, float), 'alphastd should be of type float, ' \
+        f'got {type(alphastd)} instead.'
+
+    img = img.copy().astype(np.float32)
+    if to_rgb:
+        cv2.cvtColor(img, cv2.COLOR_BGR2RGB, img)  # inplace
+
+    alpha = np.random.normal(0, alphastd, n_eigval)
+    alter = eigvec \
+        * np.broadcast_to(alpha.reshape(1, n_eigval), (3, n_eigval)) \
+        * np.broadcast_to(eigval.reshape(1, n_eigval), (3, n_eigval))
+    alter = np.broadcast_to(alter.sum(axis=1).reshape(1, 1, 3), img.shape)
+    img_adjusted = img + alter
+    return img_adjusted
+
+
+def lut_transform(img, lut_table):
+    """Transform array by look-up table.
+
+    The function lut_transform fills the output array with values from the
+    look-up table. Indices of the entries are taken from the input array.
+
+    Args:
+        img (ndarray): Image to be transformed.
+        lut_table (ndarray): look-up table of 256 elements; in case of
+            multi-channel input array, the table should either have a single
+            channel (in this case the same table is used for all channels) or
+            the same number of channels as in the input array.
+
+    Returns:
+        ndarray: The transformed image.
+    """
+    assert isinstance(img, np.ndarray)
+    assert 0 <= np.min(img) and np.max(img) <= 255
+    assert isinstance(lut_table, np.ndarray)
+    assert lut_table.shape == (256, )
+
+    return cv2.LUT(np.array(img, dtype=np.uint8), lut_table)
+
+
+def clahe(img, clip_limit=40.0, tile_grid_size=(8, 8)):
+    """Use CLAHE method to process the image.
+
+    See `ZUIDERVELD,K. Contrast Limited Adaptive Histogram Equalization[J].
+    Graphics Gems, 1994:474-485.` for more information.
+
+    Args:
+        img (ndarray): Image to be processed.
+        clip_limit (float): Threshold for contrast limiting. Default: 40.0.
+        tile_grid_size (tuple[int]): Size of grid for histogram equalization.
+            Input image will be divided into equally sized rectangular tiles.
+            It defines the number of tiles in row and column. Default: (8, 8).
+
+    Returns:
+        ndarray: The processed image.
+    """
+    assert isinstance(img, np.ndarray)
+    assert img.ndim == 2
+    assert isinstance(clip_limit, (float, int))
+    assert is_tuple_of(tile_grid_size, int)
+    assert len(tile_grid_size) == 2
+
+    clahe = cv2.createCLAHE(clip_limit, tile_grid_size)
+    return clahe.apply(np.array(img, dtype=np.uint8))
diff --git a/annotator/uniformer/mmcv/model_zoo/deprecated.json b/annotator/uniformer/mmcv/model_zoo/deprecated.json
new file mode 100644
index 0000000000000000000000000000000000000000..25cf6f28caecc22a77e3136fefa6b8dfc0e6cb5b
--- /dev/null
+++ b/annotator/uniformer/mmcv/model_zoo/deprecated.json
@@ -0,0 +1,6 @@
+{
+  "resnet50_caffe": "detectron/resnet50_caffe",
+  "resnet50_caffe_bgr": "detectron2/resnet50_caffe_bgr",
+  "resnet101_caffe": "detectron/resnet101_caffe",
+  "resnet101_caffe_bgr": "detectron2/resnet101_caffe_bgr"
+}
diff --git a/annotator/uniformer/mmcv/model_zoo/mmcls.json b/annotator/uniformer/mmcv/model_zoo/mmcls.json
new file mode 100644
index 0000000000000000000000000000000000000000..bdb311d9fe6d9f317290feedc9e37236c6cf6e8f
--- /dev/null
+++ b/annotator/uniformer/mmcv/model_zoo/mmcls.json
@@ -0,0 +1,31 @@
+{
+  "vgg11": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg11_batch256_imagenet_20210208-4271cd6c.pth",
+  "vgg13": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg13_batch256_imagenet_20210208-4d1d6080.pth",
+  "vgg16": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg16_batch256_imagenet_20210208-db26f1a5.pth",
+  "vgg19": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg19_batch256_imagenet_20210208-e6920e4a.pth",
+  "vgg11_bn": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg11_bn_batch256_imagenet_20210207-f244902c.pth",
+  "vgg13_bn": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg13_bn_batch256_imagenet_20210207-1a8b7864.pth",
+  "vgg16_bn": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg16_bn_batch256_imagenet_20210208-7e55cd29.pth",
+  "vgg19_bn": "https://download.openmmlab.com/mmclassification/v0/vgg/vgg19_bn_batch256_imagenet_20210208-da620c4f.pth",
+  "resnet18": "https://download.openmmlab.com/mmclassification/v0/resnet/resnet18_batch256_imagenet_20200708-34ab8f90.pth",
+  "resnet34": "https://download.openmmlab.com/mmclassification/v0/resnet/resnet34_batch256_imagenet_20200708-32ffb4f7.pth",
+  "resnet50": "https://download.openmmlab.com/mmclassification/v0/resnet/resnet50_batch256_imagenet_20200708-cfb998bf.pth",
+  "resnet101": "https://download.openmmlab.com/mmclassification/v0/resnet/resnet101_batch256_imagenet_20200708-753f3608.pth",
+  "resnet152": "https://download.openmmlab.com/mmclassification/v0/resnet/resnet152_batch256_imagenet_20200708-ec25b1f9.pth",
+  "resnet50_v1d": "https://download.openmmlab.com/mmclassification/v0/resnet/resnetv1d50_batch256_imagenet_20200708-1ad0ce94.pth",
+  "resnet101_v1d": "https://download.openmmlab.com/mmclassification/v0/resnet/resnetv1d101_batch256_imagenet_20200708-9cb302ef.pth",
+  "resnet152_v1d": "https://download.openmmlab.com/mmclassification/v0/resnet/resnetv1d152_batch256_imagenet_20200708-e79cb6a2.pth",
+  "resnext50_32x4d": "https://download.openmmlab.com/mmclassification/v0/resnext/resnext50_32x4d_b32x8_imagenet_20210429-56066e27.pth",
+  "resnext101_32x4d": "https://download.openmmlab.com/mmclassification/v0/resnext/resnext101_32x4d_b32x8_imagenet_20210506-e0fa3dd5.pth",
+  "resnext101_32x8d": "https://download.openmmlab.com/mmclassification/v0/resnext/resnext101_32x8d_b32x8_imagenet_20210506-23a247d5.pth",
+  "resnext152_32x4d": "https://download.openmmlab.com/mmclassification/v0/resnext/resnext152_32x4d_b32x8_imagenet_20210524-927787be.pth",
+  "se-resnet50": "https://download.openmmlab.com/mmclassification/v0/se-resnet/se-resnet50_batch256_imagenet_20200804-ae206104.pth",
+  "se-resnet101": "https://download.openmmlab.com/mmclassification/v0/se-resnet/se-resnet101_batch256_imagenet_20200804-ba5b51d4.pth",
+  "resnest50": "https://download.openmmlab.com/mmclassification/v0/resnest/resnest50_imagenet_converted-1ebf0afe.pth",
+  "resnest101": "https://download.openmmlab.com/mmclassification/v0/resnest/resnest101_imagenet_converted-032caa52.pth",
+  "resnest200": "https://download.openmmlab.com/mmclassification/v0/resnest/resnest200_imagenet_converted-581a60f2.pth",
+  "resnest269": "https://download.openmmlab.com/mmclassification/v0/resnest/resnest269_imagenet_converted-59930960.pth",
+  "shufflenet_v1": "https://download.openmmlab.com/mmclassification/v0/shufflenet_v1/shufflenet_v1_batch1024_imagenet_20200804-5d6cec73.pth",
+  "shufflenet_v2": "https://download.openmmlab.com/mmclassification/v0/shufflenet_v2/shufflenet_v2_batch1024_imagenet_20200812-5bf4721e.pth",
+  "mobilenet_v2": "https://download.openmmlab.com/mmclassification/v0/mobilenet_v2/mobilenet_v2_batch256_imagenet_20200708-3b2dc3af.pth"
+}
diff --git a/annotator/uniformer/mmcv/model_zoo/open_mmlab.json b/annotator/uniformer/mmcv/model_zoo/open_mmlab.json
new file mode 100644
index 0000000000000000000000000000000000000000..8311db4feef92faa0841c697d75efbee8430c3a0
--- /dev/null
+++ b/annotator/uniformer/mmcv/model_zoo/open_mmlab.json
@@ -0,0 +1,50 @@
+{
+  "vgg16_caffe": "https://download.openmmlab.com/pretrain/third_party/vgg16_caffe-292e1171.pth",
+  "detectron/resnet50_caffe": "https://download.openmmlab.com/pretrain/third_party/resnet50_caffe-788b5fa3.pth",
+  "detectron2/resnet50_caffe": "https://download.openmmlab.com/pretrain/third_party/resnet50_msra-5891d200.pth",
+  "detectron/resnet101_caffe": "https://download.openmmlab.com/pretrain/third_party/resnet101_caffe-3ad79236.pth",
+  "detectron2/resnet101_caffe": "https://download.openmmlab.com/pretrain/third_party/resnet101_msra-6cc46731.pth",
+  "detectron2/resnext101_32x8d": "https://download.openmmlab.com/pretrain/third_party/resnext101_32x8d-1516f1aa.pth",
+  "resnext50_32x4d": "https://download.openmmlab.com/pretrain/third_party/resnext50-32x4d-0ab1a123.pth",
+  "resnext101_32x4d": "https://download.openmmlab.com/pretrain/third_party/resnext101_32x4d-a5af3160.pth",
+  "resnext101_64x4d": "https://download.openmmlab.com/pretrain/third_party/resnext101_64x4d-ee2c6f71.pth",
+  "contrib/resnet50_gn": "https://download.openmmlab.com/pretrain/third_party/resnet50_gn_thangvubk-ad1730dd.pth",
+  "detectron/resnet50_gn": "https://download.openmmlab.com/pretrain/third_party/resnet50_gn-9186a21c.pth",
+  "detectron/resnet101_gn": "https://download.openmmlab.com/pretrain/third_party/resnet101_gn-cac0ab98.pth",
+  "jhu/resnet50_gn_ws": "https://download.openmmlab.com/pretrain/third_party/resnet50_gn_ws-15beedd8.pth",
+  "jhu/resnet101_gn_ws": "https://download.openmmlab.com/pretrain/third_party/resnet101_gn_ws-3e3c308c.pth",
+  "jhu/resnext50_32x4d_gn_ws": "https://download.openmmlab.com/pretrain/third_party/resnext50_32x4d_gn_ws-0d87ac85.pth",
+  "jhu/resnext101_32x4d_gn_ws": "https://download.openmmlab.com/pretrain/third_party/resnext101_32x4d_gn_ws-34ac1a9e.pth",
+  "jhu/resnext50_32x4d_gn": "https://download.openmmlab.com/pretrain/third_party/resnext50_32x4d_gn-c7e8b754.pth",
+  "jhu/resnext101_32x4d_gn": "https://download.openmmlab.com/pretrain/third_party/resnext101_32x4d_gn-ac3bb84e.pth",
+  "msra/hrnetv2_w18_small": "https://download.openmmlab.com/pretrain/third_party/hrnetv2_w18_small-b5a04e21.pth",
+  "msra/hrnetv2_w18": "https://download.openmmlab.com/pretrain/third_party/hrnetv2_w18-00eb2006.pth",
+  "msra/hrnetv2_w32": "https://download.openmmlab.com/pretrain/third_party/hrnetv2_w32-dc9eeb4f.pth",
+  "msra/hrnetv2_w40": "https://download.openmmlab.com/pretrain/third_party/hrnetv2_w40-ed0b031c.pth",
+  "msra/hrnetv2_w48": "https://download.openmmlab.com/pretrain/third_party/hrnetv2_w48-d2186c55.pth",
+  "bninception_caffe": "https://download.openmmlab.com/pretrain/third_party/bn_inception_caffe-ed2e8665.pth",
+  "kin400/i3d_r50_f32s2_k400": "https://download.openmmlab.com/pretrain/third_party/i3d_r50_f32s2_k400-2c57e077.pth",
+  "kin400/nl3d_r50_f32s2_k400": "https://download.openmmlab.com/pretrain/third_party/nl3d_r50_f32s2_k400-fa7e7caa.pth",
+  "res2net101_v1d_26w_4s": "https://download.openmmlab.com/pretrain/third_party/res2net101_v1d_26w_4s_mmdetv2-f0a600f9.pth",
+  "regnetx_400mf": "https://download.openmmlab.com/pretrain/third_party/regnetx_400mf-a5b10d96.pth",
+  "regnetx_800mf": "https://download.openmmlab.com/pretrain/third_party/regnetx_800mf-1f4be4c7.pth",
+  "regnetx_1.6gf": "https://download.openmmlab.com/pretrain/third_party/regnetx_1.6gf-5791c176.pth",
+  "regnetx_3.2gf": "https://download.openmmlab.com/pretrain/third_party/regnetx_3.2gf-c2599b0f.pth",
+  "regnetx_4.0gf": "https://download.openmmlab.com/pretrain/third_party/regnetx_4.0gf-a88f671e.pth",
+  "regnetx_6.4gf": "https://download.openmmlab.com/pretrain/third_party/regnetx_6.4gf-006af45d.pth",
+  "regnetx_8.0gf": "https://download.openmmlab.com/pretrain/third_party/regnetx_8.0gf-3c68abe7.pth",
+  "regnetx_12gf": "https://download.openmmlab.com/pretrain/third_party/regnetx_12gf-4c2a3350.pth",
+  "resnet18_v1c": "https://download.openmmlab.com/pretrain/third_party/resnet18_v1c-b5776b93.pth",
+  "resnet50_v1c": "https://download.openmmlab.com/pretrain/third_party/resnet50_v1c-2cccc1ad.pth",
+  "resnet101_v1c": "https://download.openmmlab.com/pretrain/third_party/resnet101_v1c-e67eebb6.pth",
+  "mmedit/vgg16": "https://download.openmmlab.com/mmediting/third_party/vgg_state_dict.pth",
+  "mmedit/res34_en_nomixup": "https://download.openmmlab.com/mmediting/third_party/model_best_resnet34_En_nomixup.pth",
+  "mmedit/mobilenet_v2": "https://download.openmmlab.com/mmediting/third_party/mobilenet_v2.pth",
+  "contrib/mobilenet_v3_large": "https://download.openmmlab.com/pretrain/third_party/mobilenet_v3_large-bc2c3fd3.pth",
+  "contrib/mobilenet_v3_small": "https://download.openmmlab.com/pretrain/third_party/mobilenet_v3_small-47085aa1.pth",
+  "resnest50": "https://download.openmmlab.com/pretrain/third_party/resnest50_d2-7497a55b.pth",
+  "resnest101": "https://download.openmmlab.com/pretrain/third_party/resnest101_d2-f3b931b2.pth",
+  "resnest200": "https://download.openmmlab.com/pretrain/third_party/resnest200_d2-ca88e41f.pth",
+  "darknet53": "https://download.openmmlab.com/pretrain/third_party/darknet53-a628ea1b.pth",
+  "mmdet/mobilenet_v2": "https://download.openmmlab.com/mmdetection/v2.0/third_party/mobilenet_v2_batch256_imagenet-ff34753d.pth"
+}
diff --git a/annotator/uniformer/mmcv/ops/__init__.py b/annotator/uniformer/mmcv/ops/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..999e090a458ee148ceca0649f1e3806a40e909bd
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/__init__.py
@@ -0,0 +1,81 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .assign_score_withk import assign_score_withk
+from .ball_query import ball_query
+from .bbox import bbox_overlaps
+from .border_align import BorderAlign, border_align
+from .box_iou_rotated import box_iou_rotated
+from .carafe import CARAFE, CARAFENaive, CARAFEPack, carafe, carafe_naive
+from .cc_attention import CrissCrossAttention
+from .contour_expand import contour_expand
+from .corner_pool import CornerPool
+from .correlation import Correlation
+from .deform_conv import DeformConv2d, DeformConv2dPack, deform_conv2d
+from .deform_roi_pool import (DeformRoIPool, DeformRoIPoolPack,
+                              ModulatedDeformRoIPoolPack, deform_roi_pool)
+from .deprecated_wrappers import Conv2d_deprecated as Conv2d
+from .deprecated_wrappers import ConvTranspose2d_deprecated as ConvTranspose2d
+from .deprecated_wrappers import Linear_deprecated as Linear
+from .deprecated_wrappers import MaxPool2d_deprecated as MaxPool2d
+from .focal_loss import (SigmoidFocalLoss, SoftmaxFocalLoss,
+                         sigmoid_focal_loss, softmax_focal_loss)
+from .furthest_point_sample import (furthest_point_sample,
+                                    furthest_point_sample_with_dist)
+from .fused_bias_leakyrelu import FusedBiasLeakyReLU, fused_bias_leakyrelu
+from .gather_points import gather_points
+from .group_points import GroupAll, QueryAndGroup, grouping_operation
+from .info import (get_compiler_version, get_compiling_cuda_version,
+                   get_onnxruntime_op_path)
+from .iou3d import boxes_iou_bev, nms_bev, nms_normal_bev
+from .knn import knn
+from .masked_conv import MaskedConv2d, masked_conv2d
+from .modulated_deform_conv import (ModulatedDeformConv2d,
+                                    ModulatedDeformConv2dPack,
+                                    modulated_deform_conv2d)
+from .multi_scale_deform_attn import MultiScaleDeformableAttention
+from .nms import batched_nms, nms, nms_match, nms_rotated, soft_nms
+from .pixel_group import pixel_group
+from .point_sample import (SimpleRoIAlign, point_sample,
+                           rel_roi_point_to_rel_img_point)
+from .points_in_boxes import (points_in_boxes_all, points_in_boxes_cpu,
+                              points_in_boxes_part)
+from .points_sampler import PointsSampler
+from .psa_mask import PSAMask
+from .roi_align import RoIAlign, roi_align
+from .roi_align_rotated import RoIAlignRotated, roi_align_rotated
+from .roi_pool import RoIPool, roi_pool
+from .roiaware_pool3d import RoIAwarePool3d
+from .roipoint_pool3d import RoIPointPool3d
+from .saconv import SAConv2d
+from .scatter_points import DynamicScatter, dynamic_scatter
+from .sync_bn import SyncBatchNorm
+from .three_interpolate import three_interpolate
+from .three_nn import three_nn
+from .tin_shift import TINShift, tin_shift
+from .upfirdn2d import upfirdn2d
+from .voxelize import Voxelization, voxelization
+
+__all__ = [
+    'bbox_overlaps', 'CARAFE', 'CARAFENaive', 'CARAFEPack', 'carafe',
+    'carafe_naive', 'CornerPool', 'DeformConv2d', 'DeformConv2dPack',
+    'deform_conv2d', 'DeformRoIPool', 'DeformRoIPoolPack',
+    'ModulatedDeformRoIPoolPack', 'deform_roi_pool', 'SigmoidFocalLoss',
+    'SoftmaxFocalLoss', 'sigmoid_focal_loss', 'softmax_focal_loss',
+    'get_compiler_version', 'get_compiling_cuda_version',
+    'get_onnxruntime_op_path', 'MaskedConv2d', 'masked_conv2d',
+    'ModulatedDeformConv2d', 'ModulatedDeformConv2dPack',
+    'modulated_deform_conv2d', 'batched_nms', 'nms', 'soft_nms', 'nms_match',
+    'RoIAlign', 'roi_align', 'RoIPool', 'roi_pool', 'SyncBatchNorm', 'Conv2d',
+    'ConvTranspose2d', 'Linear', 'MaxPool2d', 'CrissCrossAttention', 'PSAMask',
+    'point_sample', 'rel_roi_point_to_rel_img_point', 'SimpleRoIAlign',
+    'SAConv2d', 'TINShift', 'tin_shift', 'assign_score_withk',
+    'box_iou_rotated', 'RoIPointPool3d', 'nms_rotated', 'knn', 'ball_query',
+    'upfirdn2d', 'FusedBiasLeakyReLU', 'fused_bias_leakyrelu',
+    'RoIAlignRotated', 'roi_align_rotated', 'pixel_group', 'QueryAndGroup',
+    'GroupAll', 'grouping_operation', 'contour_expand', 'three_nn',
+    'three_interpolate', 'MultiScaleDeformableAttention', 'BorderAlign',
+    'border_align', 'gather_points', 'furthest_point_sample',
+    'furthest_point_sample_with_dist', 'PointsSampler', 'Correlation',
+    'boxes_iou_bev', 'nms_bev', 'nms_normal_bev', 'Voxelization',
+    'voxelization', 'dynamic_scatter', 'DynamicScatter', 'RoIAwarePool3d',
+    'points_in_boxes_part', 'points_in_boxes_cpu', 'points_in_boxes_all'
+]
diff --git a/annotator/uniformer/mmcv/ops/assign_score_withk.py b/annotator/uniformer/mmcv/ops/assign_score_withk.py
new file mode 100644
index 0000000000000000000000000000000000000000..4906adaa2cffd1b46912fbe7d4f87ef2f9fa0012
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/assign_score_withk.py
@@ -0,0 +1,123 @@
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['assign_score_withk_forward', 'assign_score_withk_backward'])
+
+
+class AssignScoreWithK(Function):
+    r"""Perform weighted sum to generate output features according to scores.
+    Modified from `PAConv <https://github.com/CVMI-Lab/PAConv/tree/main/
+    scene_seg/lib/paconv_lib/src/gpu>`_.
+
+    This is a memory-efficient CUDA implementation of assign_scores operation,
+    which first transform all point features with weight bank, then assemble
+    neighbor features with ``knn_idx`` and perform weighted sum of ``scores``.
+
+    See the `paper <https://arxiv.org/pdf/2103.14635.pdf>`_ appendix Sec. D for
+        more detailed descriptions.
+
+    Note:
+        This implementation assumes using ``neighbor`` kernel input, which is
+            (point_features - center_features, point_features).
+        See https://github.com/CVMI-Lab/PAConv/blob/main/scene_seg/model/
+        pointnet2/paconv.py#L128 for more details.
+    """
+
+    @staticmethod
+    def forward(ctx,
+                scores,
+                point_features,
+                center_features,
+                knn_idx,
+                aggregate='sum'):
+        """
+        Args:
+            scores (torch.Tensor): (B, npoint, K, M), predicted scores to
+                aggregate weight matrices in the weight bank.
+                ``npoint`` is the number of sampled centers.
+                ``K`` is the number of queried neighbors.
+                ``M`` is the number of weight matrices in the weight bank.
+            point_features (torch.Tensor): (B, N, M, out_dim)
+                Pre-computed point features to be aggregated.
+            center_features (torch.Tensor): (B, N, M, out_dim)
+                Pre-computed center features to be aggregated.
+            knn_idx (torch.Tensor): (B, npoint, K), index of sampled kNN.
+                We assume the first idx in each row is the idx of the center.
+            aggregate (str, optional): Aggregation method.
+                Can be 'sum', 'avg' or 'max'. Defaults: 'sum'.
+
+        Returns:
+            torch.Tensor: (B, out_dim, npoint, K), the aggregated features.
+        """
+        agg = {'sum': 0, 'avg': 1, 'max': 2}
+
+        B, N, M, out_dim = point_features.size()
+        _, npoint, K, _ = scores.size()
+
+        output = point_features.new_zeros((B, out_dim, npoint, K))
+        ext_module.assign_score_withk_forward(
+            point_features.contiguous(),
+            center_features.contiguous(),
+            scores.contiguous(),
+            knn_idx.contiguous(),
+            output,
+            B=B,
+            N0=N,
+            N1=npoint,
+            M=M,
+            K=K,
+            O=out_dim,
+            aggregate=agg[aggregate])
+
+        ctx.save_for_backward(output, point_features, center_features, scores,
+                              knn_idx)
+        ctx.agg = agg[aggregate]
+
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_out):
+        """
+        Args:
+            grad_out (torch.Tensor): (B, out_dim, npoint, K)
+
+        Returns:
+            grad_scores (torch.Tensor): (B, npoint, K, M)
+            grad_point_features (torch.Tensor): (B, N, M, out_dim)
+            grad_center_features (torch.Tensor): (B, N, M, out_dim)
+        """
+        _, point_features, center_features, scores, knn_idx = ctx.saved_tensors
+
+        agg = ctx.agg
+
+        B, N, M, out_dim = point_features.size()
+        _, npoint, K, _ = scores.size()
+
+        grad_point_features = point_features.new_zeros(point_features.shape)
+        grad_center_features = center_features.new_zeros(center_features.shape)
+        grad_scores = scores.new_zeros(scores.shape)
+
+        ext_module.assign_score_withk_backward(
+            grad_out.contiguous(),
+            point_features.contiguous(),
+            center_features.contiguous(),
+            scores.contiguous(),
+            knn_idx.contiguous(),
+            grad_point_features,
+            grad_center_features,
+            grad_scores,
+            B=B,
+            N0=N,
+            N1=npoint,
+            M=M,
+            K=K,
+            O=out_dim,
+            aggregate=agg)
+
+        return grad_scores, grad_point_features, \
+            grad_center_features, None, None
+
+
+assign_score_withk = AssignScoreWithK.apply
diff --git a/annotator/uniformer/mmcv/ops/ball_query.py b/annotator/uniformer/mmcv/ops/ball_query.py
new file mode 100644
index 0000000000000000000000000000000000000000..d0466847c6e5c1239e359a0397568413ebc1504a
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/ball_query.py
@@ -0,0 +1,55 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['ball_query_forward'])
+
+
+class BallQuery(Function):
+    """Find nearby points in spherical space."""
+
+    @staticmethod
+    def forward(ctx, min_radius: float, max_radius: float, sample_num: int,
+                xyz: torch.Tensor, center_xyz: torch.Tensor) -> torch.Tensor:
+        """
+        Args:
+            min_radius (float): minimum radius of the balls.
+            max_radius (float): maximum radius of the balls.
+            sample_num (int): maximum number of features in the balls.
+            xyz (Tensor): (B, N, 3) xyz coordinates of the features.
+            center_xyz (Tensor): (B, npoint, 3) centers of the ball query.
+
+        Returns:
+            Tensor: (B, npoint, nsample) tensor with the indices of
+                the features that form the query balls.
+        """
+        assert center_xyz.is_contiguous()
+        assert xyz.is_contiguous()
+        assert min_radius < max_radius
+
+        B, N, _ = xyz.size()
+        npoint = center_xyz.size(1)
+        idx = xyz.new_zeros(B, npoint, sample_num, dtype=torch.int)
+
+        ext_module.ball_query_forward(
+            center_xyz,
+            xyz,
+            idx,
+            b=B,
+            n=N,
+            m=npoint,
+            min_radius=min_radius,
+            max_radius=max_radius,
+            nsample=sample_num)
+        if torch.__version__ != 'parrots':
+            ctx.mark_non_differentiable(idx)
+        return idx
+
+    @staticmethod
+    def backward(ctx, a=None):
+        return None, None, None, None
+
+
+ball_query = BallQuery.apply
diff --git a/annotator/uniformer/mmcv/ops/bbox.py b/annotator/uniformer/mmcv/ops/bbox.py
new file mode 100644
index 0000000000000000000000000000000000000000..0c4d58b6c91f652933974f519acd3403a833e906
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/bbox.py
@@ -0,0 +1,72 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['bbox_overlaps'])
+
+
+def bbox_overlaps(bboxes1, bboxes2, mode='iou', aligned=False, offset=0):
+    """Calculate overlap between two set of bboxes.
+
+    If ``aligned`` is ``False``, then calculate the ious between each bbox
+    of bboxes1 and bboxes2, otherwise the ious between each aligned pair of
+    bboxes1 and bboxes2.
+
+    Args:
+        bboxes1 (Tensor): shape (m, 4) in <x1, y1, x2, y2> format or empty.
+        bboxes2 (Tensor): shape (n, 4) in <x1, y1, x2, y2> format or empty.
+            If aligned is ``True``, then m and n must be equal.
+        mode (str): "iou" (intersection over union) or iof (intersection over
+            foreground).
+
+    Returns:
+        ious(Tensor): shape (m, n) if aligned == False else shape (m, 1)
+
+    Example:
+        >>> bboxes1 = torch.FloatTensor([
+        >>>     [0, 0, 10, 10],
+        >>>     [10, 10, 20, 20],
+        >>>     [32, 32, 38, 42],
+        >>> ])
+        >>> bboxes2 = torch.FloatTensor([
+        >>>     [0, 0, 10, 20],
+        >>>     [0, 10, 10, 19],
+        >>>     [10, 10, 20, 20],
+        >>> ])
+        >>> bbox_overlaps(bboxes1, bboxes2)
+        tensor([[0.5000, 0.0000, 0.0000],
+                [0.0000, 0.0000, 1.0000],
+                [0.0000, 0.0000, 0.0000]])
+
+    Example:
+        >>> empty = torch.FloatTensor([])
+        >>> nonempty = torch.FloatTensor([
+        >>>     [0, 0, 10, 9],
+        >>> ])
+        >>> assert tuple(bbox_overlaps(empty, nonempty).shape) == (0, 1)
+        >>> assert tuple(bbox_overlaps(nonempty, empty).shape) == (1, 0)
+        >>> assert tuple(bbox_overlaps(empty, empty).shape) == (0, 0)
+    """
+
+    mode_dict = {'iou': 0, 'iof': 1}
+    assert mode in mode_dict.keys()
+    mode_flag = mode_dict[mode]
+    # Either the boxes are empty or the length of boxes' last dimension is 4
+    assert (bboxes1.size(-1) == 4 or bboxes1.size(0) == 0)
+    assert (bboxes2.size(-1) == 4 or bboxes2.size(0) == 0)
+    assert offset == 1 or offset == 0
+
+    rows = bboxes1.size(0)
+    cols = bboxes2.size(0)
+    if aligned:
+        assert rows == cols
+
+    if rows * cols == 0:
+        return bboxes1.new(rows, 1) if aligned else bboxes1.new(rows, cols)
+
+    if aligned:
+        ious = bboxes1.new_zeros(rows)
+    else:
+        ious = bboxes1.new_zeros((rows, cols))
+    ext_module.bbox_overlaps(
+        bboxes1, bboxes2, ious, mode=mode_flag, aligned=aligned, offset=offset)
+    return ious
diff --git a/annotator/uniformer/mmcv/ops/border_align.py b/annotator/uniformer/mmcv/ops/border_align.py
new file mode 100644
index 0000000000000000000000000000000000000000..ff305be328e9b0a15e1bbb5e6b41beb940f55c81
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/border_align.py
@@ -0,0 +1,109 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+# modified from
+# https://github.com/Megvii-BaseDetection/cvpods/blob/master/cvpods/layers/border_align.py
+
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['border_align_forward', 'border_align_backward'])
+
+
+class BorderAlignFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, boxes, pool_size):
+        return g.op(
+            'mmcv::MMCVBorderAlign', input, boxes, pool_size_i=pool_size)
+
+    @staticmethod
+    def forward(ctx, input, boxes, pool_size):
+        ctx.pool_size = pool_size
+        ctx.input_shape = input.size()
+
+        assert boxes.ndim == 3, 'boxes must be with shape [B, H*W, 4]'
+        assert boxes.size(2) == 4, \
+            'the last dimension of boxes must be (x1, y1, x2, y2)'
+        assert input.size(1) % 4 == 0, \
+            'the channel for input feature must be divisible by factor 4'
+
+        # [B, C//4, H*W, 4]
+        output_shape = (input.size(0), input.size(1) // 4, boxes.size(1), 4)
+        output = input.new_zeros(output_shape)
+        # `argmax_idx` only used for backward
+        argmax_idx = input.new_zeros(output_shape).to(torch.int)
+
+        ext_module.border_align_forward(
+            input, boxes, output, argmax_idx, pool_size=ctx.pool_size)
+
+        ctx.save_for_backward(boxes, argmax_idx)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        boxes, argmax_idx = ctx.saved_tensors
+        grad_input = grad_output.new_zeros(ctx.input_shape)
+        # complex head architecture may cause grad_output uncontiguous
+        grad_output = grad_output.contiguous()
+        ext_module.border_align_backward(
+            grad_output,
+            boxes,
+            argmax_idx,
+            grad_input,
+            pool_size=ctx.pool_size)
+        return grad_input, None, None
+
+
+border_align = BorderAlignFunction.apply
+
+
+class BorderAlign(nn.Module):
+    r"""Border align pooling layer.
+
+    Applies border_align over the input feature based on predicted bboxes.
+    The details were described in the paper
+    `BorderDet: Border Feature for Dense Object Detection
+    <https://arxiv.org/abs/2007.11056>`_.
+
+    For each border line (e.g. top, left, bottom or right) of each box,
+    border_align does the following:
+        1. uniformly samples `pool_size`+1 positions on this line, involving \
+           the start and end points.
+        2. the corresponding features on these points are computed by \
+           bilinear interpolation.
+        3. max pooling over all the `pool_size`+1 positions are used for \
+           computing pooled feature.
+
+    Args:
+        pool_size (int): number of positions sampled over the boxes' borders
+            (e.g. top, bottom, left, right).
+
+    """
+
+    def __init__(self, pool_size):
+        super(BorderAlign, self).__init__()
+        self.pool_size = pool_size
+
+    def forward(self, input, boxes):
+        """
+        Args:
+            input: Features with shape [N,4C,H,W]. Channels ranged in [0,C),
+                [C,2C), [2C,3C), [3C,4C) represent the top, left, bottom,
+                right features respectively.
+            boxes: Boxes with shape [N,H*W,4]. Coordinate format (x1,y1,x2,y2).
+
+        Returns:
+            Tensor: Pooled features with shape [N,C,H*W,4]. The order is
+                (top,left,bottom,right) for the last dimension.
+        """
+        return border_align(input, boxes, self.pool_size)
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(pool_size={self.pool_size})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/box_iou_rotated.py b/annotator/uniformer/mmcv/ops/box_iou_rotated.py
new file mode 100644
index 0000000000000000000000000000000000000000..2d78015e9c2a9e7a52859b4e18f84a9aa63481a0
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/box_iou_rotated.py
@@ -0,0 +1,45 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['box_iou_rotated'])
+
+
+def box_iou_rotated(bboxes1, bboxes2, mode='iou', aligned=False):
+    """Return intersection-over-union (Jaccard index) of boxes.
+
+    Both sets of boxes are expected to be in
+    (x_center, y_center, width, height, angle) format.
+
+    If ``aligned`` is ``False``, then calculate the ious between each bbox
+    of bboxes1 and bboxes2, otherwise the ious between each aligned pair of
+    bboxes1 and bboxes2.
+
+    Arguments:
+        boxes1 (Tensor): rotated bboxes 1. \
+            It has shape (N, 5), indicating (x, y, w, h, theta) for each row.
+            Note that theta is in radian.
+        boxes2 (Tensor): rotated bboxes 2. \
+            It has shape (M, 5), indicating (x, y, w, h, theta) for each row.
+            Note that theta is in radian.
+        mode (str): "iou" (intersection over union) or iof (intersection over
+            foreground).
+
+    Returns:
+        ious(Tensor): shape (N, M) if aligned == False else shape (N,)
+    """
+    assert mode in ['iou', 'iof']
+    mode_dict = {'iou': 0, 'iof': 1}
+    mode_flag = mode_dict[mode]
+    rows = bboxes1.size(0)
+    cols = bboxes2.size(0)
+    if aligned:
+        ious = bboxes1.new_zeros(rows)
+    else:
+        ious = bboxes1.new_zeros((rows * cols))
+    bboxes1 = bboxes1.contiguous()
+    bboxes2 = bboxes2.contiguous()
+    ext_module.box_iou_rotated(
+        bboxes1, bboxes2, ious, mode_flag=mode_flag, aligned=aligned)
+    if not aligned:
+        ious = ious.view(rows, cols)
+    return ious
diff --git a/annotator/uniformer/mmcv/ops/carafe.py b/annotator/uniformer/mmcv/ops/carafe.py
new file mode 100644
index 0000000000000000000000000000000000000000..5154cb3abfccfbbe0a1b2daa67018dbf80aaf6d2
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/carafe.py
@@ -0,0 +1,287 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.autograd import Function
+from torch.nn.modules.module import Module
+
+from ..cnn import UPSAMPLE_LAYERS, normal_init, xavier_init
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', [
+    'carafe_naive_forward', 'carafe_naive_backward', 'carafe_forward',
+    'carafe_backward'
+])
+
+
+class CARAFENaiveFunction(Function):
+
+    @staticmethod
+    def symbolic(g, features, masks, kernel_size, group_size, scale_factor):
+        return g.op(
+            'mmcv::MMCVCARAFENaive',
+            features,
+            masks,
+            kernel_size_i=kernel_size,
+            group_size_i=group_size,
+            scale_factor_f=scale_factor)
+
+    @staticmethod
+    def forward(ctx, features, masks, kernel_size, group_size, scale_factor):
+        assert scale_factor >= 1
+        assert masks.size(1) == kernel_size * kernel_size * group_size
+        assert masks.size(-1) == features.size(-1) * scale_factor
+        assert masks.size(-2) == features.size(-2) * scale_factor
+        assert features.size(1) % group_size == 0
+        assert (kernel_size - 1) % 2 == 0 and kernel_size >= 1
+        ctx.kernel_size = kernel_size
+        ctx.group_size = group_size
+        ctx.scale_factor = scale_factor
+        ctx.feature_size = features.size()
+        ctx.mask_size = masks.size()
+
+        n, c, h, w = features.size()
+        output = features.new_zeros((n, c, h * scale_factor, w * scale_factor))
+        ext_module.carafe_naive_forward(
+            features,
+            masks,
+            output,
+            kernel_size=kernel_size,
+            group_size=group_size,
+            scale_factor=scale_factor)
+
+        if features.requires_grad or masks.requires_grad:
+            ctx.save_for_backward(features, masks)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        assert grad_output.is_cuda
+
+        features, masks = ctx.saved_tensors
+        kernel_size = ctx.kernel_size
+        group_size = ctx.group_size
+        scale_factor = ctx.scale_factor
+
+        grad_input = torch.zeros_like(features)
+        grad_masks = torch.zeros_like(masks)
+        ext_module.carafe_naive_backward(
+            grad_output.contiguous(),
+            features,
+            masks,
+            grad_input,
+            grad_masks,
+            kernel_size=kernel_size,
+            group_size=group_size,
+            scale_factor=scale_factor)
+
+        return grad_input, grad_masks, None, None, None
+
+
+carafe_naive = CARAFENaiveFunction.apply
+
+
+class CARAFENaive(Module):
+
+    def __init__(self, kernel_size, group_size, scale_factor):
+        super(CARAFENaive, self).__init__()
+
+        assert isinstance(kernel_size, int) and isinstance(
+            group_size, int) and isinstance(scale_factor, int)
+        self.kernel_size = kernel_size
+        self.group_size = group_size
+        self.scale_factor = scale_factor
+
+    def forward(self, features, masks):
+        return carafe_naive(features, masks, self.kernel_size, self.group_size,
+                            self.scale_factor)
+
+
+class CARAFEFunction(Function):
+
+    @staticmethod
+    def symbolic(g, features, masks, kernel_size, group_size, scale_factor):
+        return g.op(
+            'mmcv::MMCVCARAFE',
+            features,
+            masks,
+            kernel_size_i=kernel_size,
+            group_size_i=group_size,
+            scale_factor_f=scale_factor)
+
+    @staticmethod
+    def forward(ctx, features, masks, kernel_size, group_size, scale_factor):
+        assert scale_factor >= 1
+        assert masks.size(1) == kernel_size * kernel_size * group_size
+        assert masks.size(-1) == features.size(-1) * scale_factor
+        assert masks.size(-2) == features.size(-2) * scale_factor
+        assert features.size(1) % group_size == 0
+        assert (kernel_size - 1) % 2 == 0 and kernel_size >= 1
+        ctx.kernel_size = kernel_size
+        ctx.group_size = group_size
+        ctx.scale_factor = scale_factor
+        ctx.feature_size = features.size()
+        ctx.mask_size = masks.size()
+
+        n, c, h, w = features.size()
+        output = features.new_zeros((n, c, h * scale_factor, w * scale_factor))
+        routput = features.new_zeros(output.size(), requires_grad=False)
+        rfeatures = features.new_zeros(features.size(), requires_grad=False)
+        rmasks = masks.new_zeros(masks.size(), requires_grad=False)
+        ext_module.carafe_forward(
+            features,
+            masks,
+            rfeatures,
+            routput,
+            rmasks,
+            output,
+            kernel_size=kernel_size,
+            group_size=group_size,
+            scale_factor=scale_factor)
+
+        if features.requires_grad or masks.requires_grad:
+            ctx.save_for_backward(features, masks, rfeatures)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        assert grad_output.is_cuda
+
+        features, masks, rfeatures = ctx.saved_tensors
+        kernel_size = ctx.kernel_size
+        group_size = ctx.group_size
+        scale_factor = ctx.scale_factor
+
+        rgrad_output = torch.zeros_like(grad_output, requires_grad=False)
+        rgrad_input_hs = torch.zeros_like(grad_output, requires_grad=False)
+        rgrad_input = torch.zeros_like(features, requires_grad=False)
+        rgrad_masks = torch.zeros_like(masks, requires_grad=False)
+        grad_input = torch.zeros_like(features, requires_grad=False)
+        grad_masks = torch.zeros_like(masks, requires_grad=False)
+        ext_module.carafe_backward(
+            grad_output.contiguous(),
+            rfeatures,
+            masks,
+            rgrad_output,
+            rgrad_input_hs,
+            rgrad_input,
+            rgrad_masks,
+            grad_input,
+            grad_masks,
+            kernel_size=kernel_size,
+            group_size=group_size,
+            scale_factor=scale_factor)
+        return grad_input, grad_masks, None, None, None
+
+
+carafe = CARAFEFunction.apply
+
+
+class CARAFE(Module):
+    """ CARAFE: Content-Aware ReAssembly of FEatures
+
+    Please refer to https://arxiv.org/abs/1905.02188 for more details.
+
+    Args:
+        kernel_size (int): reassemble kernel size
+        group_size (int): reassemble group size
+        scale_factor (int): upsample ratio
+
+    Returns:
+        upsampled feature map
+    """
+
+    def __init__(self, kernel_size, group_size, scale_factor):
+        super(CARAFE, self).__init__()
+
+        assert isinstance(kernel_size, int) and isinstance(
+            group_size, int) and isinstance(scale_factor, int)
+        self.kernel_size = kernel_size
+        self.group_size = group_size
+        self.scale_factor = scale_factor
+
+    def forward(self, features, masks):
+        return carafe(features, masks, self.kernel_size, self.group_size,
+                      self.scale_factor)
+
+
+@UPSAMPLE_LAYERS.register_module(name='carafe')
+class CARAFEPack(nn.Module):
+    """A unified package of CARAFE upsampler that contains: 1) channel
+    compressor 2) content encoder 3) CARAFE op.
+
+    Official implementation of ICCV 2019 paper
+    CARAFE: Content-Aware ReAssembly of FEatures
+    Please refer to https://arxiv.org/abs/1905.02188 for more details.
+
+    Args:
+        channels (int): input feature channels
+        scale_factor (int): upsample ratio
+        up_kernel (int): kernel size of CARAFE op
+        up_group (int): group size of CARAFE op
+        encoder_kernel (int): kernel size of content encoder
+        encoder_dilation (int): dilation of content encoder
+        compressed_channels (int): output channels of channels compressor
+
+    Returns:
+        upsampled feature map
+    """
+
+    def __init__(self,
+                 channels,
+                 scale_factor,
+                 up_kernel=5,
+                 up_group=1,
+                 encoder_kernel=3,
+                 encoder_dilation=1,
+                 compressed_channels=64):
+        super(CARAFEPack, self).__init__()
+        self.channels = channels
+        self.scale_factor = scale_factor
+        self.up_kernel = up_kernel
+        self.up_group = up_group
+        self.encoder_kernel = encoder_kernel
+        self.encoder_dilation = encoder_dilation
+        self.compressed_channels = compressed_channels
+        self.channel_compressor = nn.Conv2d(channels, self.compressed_channels,
+                                            1)
+        self.content_encoder = nn.Conv2d(
+            self.compressed_channels,
+            self.up_kernel * self.up_kernel * self.up_group *
+            self.scale_factor * self.scale_factor,
+            self.encoder_kernel,
+            padding=int((self.encoder_kernel - 1) * self.encoder_dilation / 2),
+            dilation=self.encoder_dilation,
+            groups=1)
+        self.init_weights()
+
+    def init_weights(self):
+        for m in self.modules():
+            if isinstance(m, nn.Conv2d):
+                xavier_init(m, distribution='uniform')
+        normal_init(self.content_encoder, std=0.001)
+
+    def kernel_normalizer(self, mask):
+        mask = F.pixel_shuffle(mask, self.scale_factor)
+        n, mask_c, h, w = mask.size()
+        # use float division explicitly,
+        # to void inconsistency while exporting to onnx
+        mask_channel = int(mask_c / float(self.up_kernel**2))
+        mask = mask.view(n, mask_channel, -1, h, w)
+
+        mask = F.softmax(mask, dim=2, dtype=mask.dtype)
+        mask = mask.view(n, mask_c, h, w).contiguous()
+
+        return mask
+
+    def feature_reassemble(self, x, mask):
+        x = carafe(x, mask, self.up_kernel, self.up_group, self.scale_factor)
+        return x
+
+    def forward(self, x):
+        compressed_x = self.channel_compressor(x)
+        mask = self.content_encoder(compressed_x)
+        mask = self.kernel_normalizer(mask)
+
+        x = self.feature_reassemble(x, mask)
+        return x
diff --git a/annotator/uniformer/mmcv/ops/cc_attention.py b/annotator/uniformer/mmcv/ops/cc_attention.py
new file mode 100644
index 0000000000000000000000000000000000000000..9207aa95e6730bd9b3362dee612059a5f0ce1c5e
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/cc_attention.py
@@ -0,0 +1,83 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from annotator.uniformer.mmcv.cnn import PLUGIN_LAYERS, Scale
+
+
+def NEG_INF_DIAG(n, device):
+    """Returns a diagonal matrix of size [n, n].
+
+    The diagonal are all "-inf". This is for avoiding calculating the
+    overlapped element in the Criss-Cross twice.
+    """
+    return torch.diag(torch.tensor(float('-inf')).to(device).repeat(n), 0)
+
+
+@PLUGIN_LAYERS.register_module()
+class CrissCrossAttention(nn.Module):
+    """Criss-Cross Attention Module.
+
+    .. note::
+        Before v1.3.13, we use a CUDA op. Since v1.3.13, we switch
+        to a pure PyTorch and equivalent implementation. For more
+        details, please refer to https://github.com/open-mmlab/mmcv/pull/1201.
+
+        Speed comparison for one forward pass
+
+        - Input size: [2,512,97,97]
+        - Device: 1 NVIDIA GeForce RTX 2080 Ti
+
+        +-----------------------+---------------+------------+---------------+
+        |                       |PyTorch version|CUDA version|Relative speed |
+        +=======================+===============+============+===============+
+        |with torch.no_grad()   |0.00554402 s   |0.0299619 s |5.4x           |
+        +-----------------------+---------------+------------+---------------+
+        |no with torch.no_grad()|0.00562803 s   |0.0301349 s |5.4x           |
+        +-----------------------+---------------+------------+---------------+
+
+    Args:
+        in_channels (int): Channels of the input feature map.
+    """
+
+    def __init__(self, in_channels):
+        super().__init__()
+        self.query_conv = nn.Conv2d(in_channels, in_channels // 8, 1)
+        self.key_conv = nn.Conv2d(in_channels, in_channels // 8, 1)
+        self.value_conv = nn.Conv2d(in_channels, in_channels, 1)
+        self.gamma = Scale(0.)
+        self.in_channels = in_channels
+
+    def forward(self, x):
+        """forward function of Criss-Cross Attention.
+
+        Args:
+            x (Tensor): Input feature. \
+                shape (batch_size, in_channels, height, width)
+        Returns:
+            Tensor: Output of the layer, with shape of \
+            (batch_size, in_channels, height, width)
+        """
+        B, C, H, W = x.size()
+        query = self.query_conv(x)
+        key = self.key_conv(x)
+        value = self.value_conv(x)
+        energy_H = torch.einsum('bchw,bciw->bwhi', query, key) + NEG_INF_DIAG(
+            H, query.device)
+        energy_H = energy_H.transpose(1, 2)
+        energy_W = torch.einsum('bchw,bchj->bhwj', query, key)
+        attn = F.softmax(
+            torch.cat([energy_H, energy_W], dim=-1), dim=-1)  # [B,H,W,(H+W)]
+        out = torch.einsum('bciw,bhwi->bchw', value, attn[..., :H])
+        out += torch.einsum('bchj,bhwj->bchw', value, attn[..., H:])
+
+        out = self.gamma(out) + x
+        out = out.contiguous()
+
+        return out
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(in_channels={self.in_channels})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/contour_expand.py b/annotator/uniformer/mmcv/ops/contour_expand.py
new file mode 100644
index 0000000000000000000000000000000000000000..ea1111e1768b5f27e118bf7dbc0d9c70a7afd6d7
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/contour_expand.py
@@ -0,0 +1,49 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import numpy as np
+import torch
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['contour_expand'])
+
+
+def contour_expand(kernel_mask, internal_kernel_label, min_kernel_area,
+                   kernel_num):
+    """Expand kernel contours so that foreground pixels are assigned into
+    instances.
+
+    Arguments:
+        kernel_mask (np.array or Tensor): The instance kernel mask with
+            size hxw.
+        internal_kernel_label (np.array or Tensor): The instance internal
+            kernel label with size hxw.
+        min_kernel_area (int): The minimum kernel area.
+        kernel_num (int): The instance kernel number.
+
+    Returns:
+        label (list): The instance index map with size hxw.
+    """
+    assert isinstance(kernel_mask, (torch.Tensor, np.ndarray))
+    assert isinstance(internal_kernel_label, (torch.Tensor, np.ndarray))
+    assert isinstance(min_kernel_area, int)
+    assert isinstance(kernel_num, int)
+
+    if isinstance(kernel_mask, np.ndarray):
+        kernel_mask = torch.from_numpy(kernel_mask)
+    if isinstance(internal_kernel_label, np.ndarray):
+        internal_kernel_label = torch.from_numpy(internal_kernel_label)
+
+    if torch.__version__ == 'parrots':
+        if kernel_mask.shape[0] == 0 or internal_kernel_label.shape[0] == 0:
+            label = []
+        else:
+            label = ext_module.contour_expand(
+                kernel_mask,
+                internal_kernel_label,
+                min_kernel_area=min_kernel_area,
+                kernel_num=kernel_num)
+            label = label.tolist()
+    else:
+        label = ext_module.contour_expand(kernel_mask, internal_kernel_label,
+                                          min_kernel_area, kernel_num)
+    return label
diff --git a/annotator/uniformer/mmcv/ops/corner_pool.py b/annotator/uniformer/mmcv/ops/corner_pool.py
new file mode 100644
index 0000000000000000000000000000000000000000..a33d798b43d405e4c86bee4cd6389be21ca9c637
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/corner_pool.py
@@ -0,0 +1,161 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch import nn
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', [
+    'top_pool_forward', 'top_pool_backward', 'bottom_pool_forward',
+    'bottom_pool_backward', 'left_pool_forward', 'left_pool_backward',
+    'right_pool_forward', 'right_pool_backward'
+])
+
+_mode_dict = {'top': 0, 'bottom': 1, 'left': 2, 'right': 3}
+
+
+class TopPoolFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input):
+        output = g.op(
+            'mmcv::MMCVCornerPool', input, mode_i=int(_mode_dict['top']))
+        return output
+
+    @staticmethod
+    def forward(ctx, input):
+        output = ext_module.top_pool_forward(input)
+        ctx.save_for_backward(input)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        input, = ctx.saved_tensors
+        output = ext_module.top_pool_backward(input, grad_output)
+        return output
+
+
+class BottomPoolFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input):
+        output = g.op(
+            'mmcv::MMCVCornerPool', input, mode_i=int(_mode_dict['bottom']))
+        return output
+
+    @staticmethod
+    def forward(ctx, input):
+        output = ext_module.bottom_pool_forward(input)
+        ctx.save_for_backward(input)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        input, = ctx.saved_tensors
+        output = ext_module.bottom_pool_backward(input, grad_output)
+        return output
+
+
+class LeftPoolFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input):
+        output = g.op(
+            'mmcv::MMCVCornerPool', input, mode_i=int(_mode_dict['left']))
+        return output
+
+    @staticmethod
+    def forward(ctx, input):
+        output = ext_module.left_pool_forward(input)
+        ctx.save_for_backward(input)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        input, = ctx.saved_tensors
+        output = ext_module.left_pool_backward(input, grad_output)
+        return output
+
+
+class RightPoolFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input):
+        output = g.op(
+            'mmcv::MMCVCornerPool', input, mode_i=int(_mode_dict['right']))
+        return output
+
+    @staticmethod
+    def forward(ctx, input):
+        output = ext_module.right_pool_forward(input)
+        ctx.save_for_backward(input)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        input, = ctx.saved_tensors
+        output = ext_module.right_pool_backward(input, grad_output)
+        return output
+
+
+class CornerPool(nn.Module):
+    """Corner Pooling.
+
+    Corner Pooling is a new type of pooling layer that helps a
+    convolutional network better localize corners of bounding boxes.
+
+    Please refer to https://arxiv.org/abs/1808.01244 for more details.
+    Code is modified from https://github.com/princeton-vl/CornerNet-Lite.
+
+    Args:
+        mode(str): Pooling orientation for the pooling layer
+
+            - 'bottom': Bottom Pooling
+            - 'left': Left Pooling
+            - 'right': Right Pooling
+            - 'top': Top Pooling
+
+    Returns:
+        Feature map after pooling.
+    """
+
+    pool_functions = {
+        'bottom': BottomPoolFunction,
+        'left': LeftPoolFunction,
+        'right': RightPoolFunction,
+        'top': TopPoolFunction,
+    }
+
+    cummax_dim_flip = {
+        'bottom': (2, False),
+        'left': (3, True),
+        'right': (3, False),
+        'top': (2, True),
+    }
+
+    def __init__(self, mode):
+        super(CornerPool, self).__init__()
+        assert mode in self.pool_functions
+        self.mode = mode
+        self.corner_pool = self.pool_functions[mode]
+
+    def forward(self, x):
+        if torch.__version__ != 'parrots' and torch.__version__ >= '1.5.0':
+            if torch.onnx.is_in_onnx_export():
+                assert torch.__version__ >= '1.7.0', \
+                    'When `cummax` serves as an intermediate component whose '\
+                    'outputs is used as inputs for another modules, it\'s '\
+                    'expected that pytorch version must be >= 1.7.0, '\
+                    'otherwise Error appears like: `RuntimeError: tuple '\
+                    'appears in op that does not forward tuples, unsupported '\
+                    'kind: prim::PythonOp`.'
+
+            dim, flip = self.cummax_dim_flip[self.mode]
+            if flip:
+                x = x.flip(dim)
+            pool_tensor, _ = torch.cummax(x, dim=dim)
+            if flip:
+                pool_tensor = pool_tensor.flip(dim)
+            return pool_tensor
+        else:
+            return self.corner_pool.apply(x)
diff --git a/annotator/uniformer/mmcv/ops/correlation.py b/annotator/uniformer/mmcv/ops/correlation.py
new file mode 100644
index 0000000000000000000000000000000000000000..3d0b79c301b29915dfaf4d2b1846c59be73127d3
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/correlation.py
@@ -0,0 +1,196 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch import Tensor, nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.utils import _pair
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['correlation_forward', 'correlation_backward'])
+
+
+class CorrelationFunction(Function):
+
+    @staticmethod
+    def forward(ctx,
+                input1,
+                input2,
+                kernel_size=1,
+                max_displacement=1,
+                stride=1,
+                padding=1,
+                dilation=1,
+                dilation_patch=1):
+
+        ctx.save_for_backward(input1, input2)
+
+        kH, kW = ctx.kernel_size = _pair(kernel_size)
+        patch_size = max_displacement * 2 + 1
+        ctx.patch_size = patch_size
+        dH, dW = ctx.stride = _pair(stride)
+        padH, padW = ctx.padding = _pair(padding)
+        dilationH, dilationW = ctx.dilation = _pair(dilation)
+        dilation_patchH, dilation_patchW = ctx.dilation_patch = _pair(
+            dilation_patch)
+
+        output_size = CorrelationFunction._output_size(ctx, input1)
+
+        output = input1.new_zeros(output_size)
+
+        ext_module.correlation_forward(
+            input1,
+            input2,
+            output,
+            kH=kH,
+            kW=kW,
+            patchH=patch_size,
+            patchW=patch_size,
+            padH=padH,
+            padW=padW,
+            dilationH=dilationH,
+            dilationW=dilationW,
+            dilation_patchH=dilation_patchH,
+            dilation_patchW=dilation_patchW,
+            dH=dH,
+            dW=dW)
+
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        input1, input2 = ctx.saved_tensors
+
+        kH, kW = ctx.kernel_size
+        patch_size = ctx.patch_size
+        padH, padW = ctx.padding
+        dilationH, dilationW = ctx.dilation
+        dilation_patchH, dilation_patchW = ctx.dilation_patch
+        dH, dW = ctx.stride
+        grad_input1 = torch.zeros_like(input1)
+        grad_input2 = torch.zeros_like(input2)
+
+        ext_module.correlation_backward(
+            grad_output,
+            input1,
+            input2,
+            grad_input1,
+            grad_input2,
+            kH=kH,
+            kW=kW,
+            patchH=patch_size,
+            patchW=patch_size,
+            padH=padH,
+            padW=padW,
+            dilationH=dilationH,
+            dilationW=dilationW,
+            dilation_patchH=dilation_patchH,
+            dilation_patchW=dilation_patchW,
+            dH=dH,
+            dW=dW)
+        return grad_input1, grad_input2, None, None, None, None, None, None
+
+    @staticmethod
+    def _output_size(ctx, input1):
+        iH, iW = input1.size(2), input1.size(3)
+        batch_size = input1.size(0)
+        kH, kW = ctx.kernel_size
+        patch_size = ctx.patch_size
+        dH, dW = ctx.stride
+        padH, padW = ctx.padding
+        dilationH, dilationW = ctx.dilation
+        dilatedKH = (kH - 1) * dilationH + 1
+        dilatedKW = (kW - 1) * dilationW + 1
+
+        oH = int((iH + 2 * padH - dilatedKH) / dH + 1)
+        oW = int((iW + 2 * padW - dilatedKW) / dW + 1)
+
+        output_size = (batch_size, patch_size, patch_size, oH, oW)
+        return output_size
+
+
+class Correlation(nn.Module):
+    r"""Correlation operator
+
+    This correlation operator works for optical flow correlation computation.
+
+    There are two batched tensors with shape :math:`(N, C, H, W)`,
+    and the correlation output's shape is :math:`(N, max\_displacement \times
+    2 + 1, max\_displacement * 2 + 1, H_{out}, W_{out})`
+
+    where
+
+    .. math::
+        H_{out} = \left\lfloor\frac{H_{in}  + 2 \times padding -
+            dilation \times (kernel\_size - 1) - 1}
+            {stride} + 1\right\rfloor
+
+    .. math::
+        W_{out} = \left\lfloor\frac{W_{in}  + 2 \times padding - dilation
+            \times (kernel\_size - 1) - 1}
+            {stride} + 1\right\rfloor
+
+    the correlation item :math:`(N_i, dy, dx)` is formed by taking the sliding
+    window convolution between input1 and shifted input2,
+
+    .. math::
+        Corr(N_i, dx, dy) =
+        \sum_{c=0}^{C-1}
+        input1(N_i, c) \star
+        \mathcal{S}(input2(N_i, c), dy, dx)
+
+    where :math:`\star` is the valid 2d sliding window convolution operator,
+    and :math:`\mathcal{S}` means shifting the input features (auto-complete
+    zero marginal), and :math:`dx, dy` are shifting distance, :math:`dx, dy \in
+    [-max\_displacement \times dilation\_patch, max\_displacement \times
+    dilation\_patch]`.
+
+    Args:
+        kernel_size (int): The size of sliding window i.e. local neighborhood
+            representing the center points and involved in correlation
+            computation. Defaults to 1.
+        max_displacement (int): The radius for computing correlation volume,
+            but the actual working space can be dilated by dilation_patch.
+            Defaults to 1.
+        stride (int): The stride of the sliding blocks in the input spatial
+            dimensions. Defaults to 1.
+        padding (int): Zero padding added to all four sides of the input1.
+            Defaults to 0.
+        dilation (int): The spacing of local neighborhood that will involved
+            in correlation. Defaults to 1.
+        dilation_patch (int): The spacing between position need to compute
+            correlation.  Defaults to 1.
+    """
+
+    def __init__(self,
+                 kernel_size: int = 1,
+                 max_displacement: int = 1,
+                 stride: int = 1,
+                 padding: int = 0,
+                 dilation: int = 1,
+                 dilation_patch: int = 1) -> None:
+        super().__init__()
+        self.kernel_size = kernel_size
+        self.max_displacement = max_displacement
+        self.stride = stride
+        self.padding = padding
+        self.dilation = dilation
+        self.dilation_patch = dilation_patch
+
+    def forward(self, input1: Tensor, input2: Tensor) -> Tensor:
+        return CorrelationFunction.apply(input1, input2, self.kernel_size,
+                                         self.max_displacement, self.stride,
+                                         self.padding, self.dilation,
+                                         self.dilation_patch)
+
+    def __repr__(self) -> str:
+        s = self.__class__.__name__
+        s += f'(kernel_size={self.kernel_size}, '
+        s += f'max_displacement={self.max_displacement}, '
+        s += f'stride={self.stride}, '
+        s += f'padding={self.padding}, '
+        s += f'dilation={self.dilation}, '
+        s += f'dilation_patch={self.dilation_patch})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/deform_conv.py b/annotator/uniformer/mmcv/ops/deform_conv.py
new file mode 100644
index 0000000000000000000000000000000000000000..a3f8c75ee774823eea334e3b3732af6a18f55038
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/deform_conv.py
@@ -0,0 +1,405 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from typing import Tuple, Union
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch import Tensor
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.utils import _pair, _single
+
+from annotator.uniformer.mmcv.utils import deprecated_api_warning
+from ..cnn import CONV_LAYERS
+from ..utils import ext_loader, print_log
+
+ext_module = ext_loader.load_ext('_ext', [
+    'deform_conv_forward', 'deform_conv_backward_input',
+    'deform_conv_backward_parameters'
+])
+
+
+class DeformConv2dFunction(Function):
+
+    @staticmethod
+    def symbolic(g,
+                 input,
+                 offset,
+                 weight,
+                 stride,
+                 padding,
+                 dilation,
+                 groups,
+                 deform_groups,
+                 bias=False,
+                 im2col_step=32):
+        return g.op(
+            'mmcv::MMCVDeformConv2d',
+            input,
+            offset,
+            weight,
+            stride_i=stride,
+            padding_i=padding,
+            dilation_i=dilation,
+            groups_i=groups,
+            deform_groups_i=deform_groups,
+            bias_i=bias,
+            im2col_step_i=im2col_step)
+
+    @staticmethod
+    def forward(ctx,
+                input,
+                offset,
+                weight,
+                stride=1,
+                padding=0,
+                dilation=1,
+                groups=1,
+                deform_groups=1,
+                bias=False,
+                im2col_step=32):
+        if input is not None and input.dim() != 4:
+            raise ValueError(
+                f'Expected 4D tensor as input, got {input.dim()}D tensor \
+                  instead.')
+        assert bias is False, 'Only support bias is False.'
+        ctx.stride = _pair(stride)
+        ctx.padding = _pair(padding)
+        ctx.dilation = _pair(dilation)
+        ctx.groups = groups
+        ctx.deform_groups = deform_groups
+        ctx.im2col_step = im2col_step
+
+        # When pytorch version >= 1.6.0, amp is adopted for fp16 mode;
+        # amp won't cast the type of model (float32), but "offset" is cast
+        # to float16 by nn.Conv2d automatically, leading to the type
+        # mismatch with input (when it is float32) or weight.
+        # The flag for whether to use fp16 or amp is the type of "offset",
+        # we cast weight and input to temporarily support fp16 and amp
+        # whatever the pytorch version is.
+        input = input.type_as(offset)
+        weight = weight.type_as(input)
+        ctx.save_for_backward(input, offset, weight)
+
+        output = input.new_empty(
+            DeformConv2dFunction._output_size(ctx, input, weight))
+
+        ctx.bufs_ = [input.new_empty(0), input.new_empty(0)]  # columns, ones
+
+        cur_im2col_step = min(ctx.im2col_step, input.size(0))
+        assert (input.size(0) %
+                cur_im2col_step) == 0, 'im2col step must divide batchsize'
+        ext_module.deform_conv_forward(
+            input,
+            weight,
+            offset,
+            output,
+            ctx.bufs_[0],
+            ctx.bufs_[1],
+            kW=weight.size(3),
+            kH=weight.size(2),
+            dW=ctx.stride[1],
+            dH=ctx.stride[0],
+            padW=ctx.padding[1],
+            padH=ctx.padding[0],
+            dilationW=ctx.dilation[1],
+            dilationH=ctx.dilation[0],
+            group=ctx.groups,
+            deformable_group=ctx.deform_groups,
+            im2col_step=cur_im2col_step)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        input, offset, weight = ctx.saved_tensors
+
+        grad_input = grad_offset = grad_weight = None
+
+        cur_im2col_step = min(ctx.im2col_step, input.size(0))
+        assert (input.size(0) % cur_im2col_step
+                ) == 0, 'batch size must be divisible by im2col_step'
+
+        grad_output = grad_output.contiguous()
+        if ctx.needs_input_grad[0] or ctx.needs_input_grad[1]:
+            grad_input = torch.zeros_like(input)
+            grad_offset = torch.zeros_like(offset)
+            ext_module.deform_conv_backward_input(
+                input,
+                offset,
+                grad_output,
+                grad_input,
+                grad_offset,
+                weight,
+                ctx.bufs_[0],
+                kW=weight.size(3),
+                kH=weight.size(2),
+                dW=ctx.stride[1],
+                dH=ctx.stride[0],
+                padW=ctx.padding[1],
+                padH=ctx.padding[0],
+                dilationW=ctx.dilation[1],
+                dilationH=ctx.dilation[0],
+                group=ctx.groups,
+                deformable_group=ctx.deform_groups,
+                im2col_step=cur_im2col_step)
+
+        if ctx.needs_input_grad[2]:
+            grad_weight = torch.zeros_like(weight)
+            ext_module.deform_conv_backward_parameters(
+                input,
+                offset,
+                grad_output,
+                grad_weight,
+                ctx.bufs_[0],
+                ctx.bufs_[1],
+                kW=weight.size(3),
+                kH=weight.size(2),
+                dW=ctx.stride[1],
+                dH=ctx.stride[0],
+                padW=ctx.padding[1],
+                padH=ctx.padding[0],
+                dilationW=ctx.dilation[1],
+                dilationH=ctx.dilation[0],
+                group=ctx.groups,
+                deformable_group=ctx.deform_groups,
+                scale=1,
+                im2col_step=cur_im2col_step)
+
+        return grad_input, grad_offset, grad_weight, \
+            None, None, None, None, None, None, None
+
+    @staticmethod
+    def _output_size(ctx, input, weight):
+        channels = weight.size(0)
+        output_size = (input.size(0), channels)
+        for d in range(input.dim() - 2):
+            in_size = input.size(d + 2)
+            pad = ctx.padding[d]
+            kernel = ctx.dilation[d] * (weight.size(d + 2) - 1) + 1
+            stride_ = ctx.stride[d]
+            output_size += ((in_size + (2 * pad) - kernel) // stride_ + 1, )
+        if not all(map(lambda s: s > 0, output_size)):
+            raise ValueError(
+                'convolution input is too small (output would be ' +
+                'x'.join(map(str, output_size)) + ')')
+        return output_size
+
+
+deform_conv2d = DeformConv2dFunction.apply
+
+
+class DeformConv2d(nn.Module):
+    r"""Deformable 2D convolution.
+
+    Applies a deformable 2D convolution over an input signal composed of
+    several input planes. DeformConv2d was described in the paper
+    `Deformable Convolutional Networks
+    <https://arxiv.org/pdf/1703.06211.pdf>`_
+
+    Note:
+        The argument ``im2col_step`` was added in version 1.3.17, which means
+        number of samples processed by the ``im2col_cuda_kernel`` per call.
+        It enables users to define ``batch_size`` and ``im2col_step`` more
+        flexibly and solved `issue mmcv#1440
+        <https://github.com/open-mmlab/mmcv/issues/1440>`_.
+
+    Args:
+        in_channels (int): Number of channels in the input image.
+        out_channels (int): Number of channels produced by the convolution.
+        kernel_size(int, tuple): Size of the convolving kernel.
+        stride(int, tuple): Stride of the convolution. Default: 1.
+        padding (int or tuple): Zero-padding added to both sides of the input.
+            Default: 0.
+        dilation (int or tuple): Spacing between kernel elements. Default: 1.
+        groups (int): Number of blocked connections from input.
+            channels to output channels. Default: 1.
+        deform_groups (int): Number of deformable group partitions.
+        bias (bool): If True, adds a learnable bias to the output.
+            Default: False.
+        im2col_step (int): Number of samples processed by im2col_cuda_kernel
+            per call. It will work when ``batch_size`` > ``im2col_step``, but
+            ``batch_size`` must be divisible by ``im2col_step``. Default: 32.
+            `New in version 1.3.17.`
+    """
+
+    @deprecated_api_warning({'deformable_groups': 'deform_groups'},
+                            cls_name='DeformConv2d')
+    def __init__(self,
+                 in_channels: int,
+                 out_channels: int,
+                 kernel_size: Union[int, Tuple[int, ...]],
+                 stride: Union[int, Tuple[int, ...]] = 1,
+                 padding: Union[int, Tuple[int, ...]] = 0,
+                 dilation: Union[int, Tuple[int, ...]] = 1,
+                 groups: int = 1,
+                 deform_groups: int = 1,
+                 bias: bool = False,
+                 im2col_step: int = 32) -> None:
+        super(DeformConv2d, self).__init__()
+
+        assert not bias, \
+            f'bias={bias} is not supported in DeformConv2d.'
+        assert in_channels % groups == 0, \
+            f'in_channels {in_channels} cannot be divisible by groups {groups}'
+        assert out_channels % groups == 0, \
+            f'out_channels {out_channels} cannot be divisible by groups \
+              {groups}'
+
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.kernel_size = _pair(kernel_size)
+        self.stride = _pair(stride)
+        self.padding = _pair(padding)
+        self.dilation = _pair(dilation)
+        self.groups = groups
+        self.deform_groups = deform_groups
+        self.im2col_step = im2col_step
+        # enable compatibility with nn.Conv2d
+        self.transposed = False
+        self.output_padding = _single(0)
+
+        # only weight, no bias
+        self.weight = nn.Parameter(
+            torch.Tensor(out_channels, in_channels // self.groups,
+                         *self.kernel_size))
+
+        self.reset_parameters()
+
+    def reset_parameters(self):
+        # switch the initialization of `self.weight` to the standard kaiming
+        # method described in `Delving deep into rectifiers: Surpassing
+        # human-level performance on ImageNet classification` - He, K. et al.
+        # (2015), using a uniform distribution
+        nn.init.kaiming_uniform_(self.weight, nonlinearity='relu')
+
+    def forward(self, x: Tensor, offset: Tensor) -> Tensor:
+        """Deformable Convolutional forward function.
+
+        Args:
+            x (Tensor): Input feature, shape (B, C_in, H_in, W_in)
+            offset (Tensor): Offset for deformable convolution, shape
+                (B, deform_groups*kernel_size[0]*kernel_size[1]*2,
+                H_out, W_out), H_out, W_out are equal to the output's.
+
+                An offset is like `[y0, x0, y1, x1, y2, x2, ..., y8, x8]`.
+                The spatial arrangement is like:
+
+                .. code:: text
+
+                    (x0, y0) (x1, y1) (x2, y2)
+                    (x3, y3) (x4, y4) (x5, y5)
+                    (x6, y6) (x7, y7) (x8, y8)
+
+        Returns:
+            Tensor: Output of the layer.
+        """
+        # To fix an assert error in deform_conv_cuda.cpp:128
+        # input image is smaller than kernel
+        input_pad = (x.size(2) < self.kernel_size[0]) or (x.size(3) <
+                                                          self.kernel_size[1])
+        if input_pad:
+            pad_h = max(self.kernel_size[0] - x.size(2), 0)
+            pad_w = max(self.kernel_size[1] - x.size(3), 0)
+            x = F.pad(x, (0, pad_w, 0, pad_h), 'constant', 0).contiguous()
+            offset = F.pad(offset, (0, pad_w, 0, pad_h), 'constant', 0)
+            offset = offset.contiguous()
+        out = deform_conv2d(x, offset, self.weight, self.stride, self.padding,
+                            self.dilation, self.groups, self.deform_groups,
+                            False, self.im2col_step)
+        if input_pad:
+            out = out[:, :, :out.size(2) - pad_h, :out.size(3) -
+                      pad_w].contiguous()
+        return out
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(in_channels={self.in_channels},\n'
+        s += f'out_channels={self.out_channels},\n'
+        s += f'kernel_size={self.kernel_size},\n'
+        s += f'stride={self.stride},\n'
+        s += f'padding={self.padding},\n'
+        s += f'dilation={self.dilation},\n'
+        s += f'groups={self.groups},\n'
+        s += f'deform_groups={self.deform_groups},\n'
+        # bias is not supported in DeformConv2d.
+        s += 'bias=False)'
+        return s
+
+
+@CONV_LAYERS.register_module('DCN')
+class DeformConv2dPack(DeformConv2d):
+    """A Deformable Conv Encapsulation that acts as normal Conv layers.
+
+    The offset tensor is like `[y0, x0, y1, x1, y2, x2, ..., y8, x8]`.
+    The spatial arrangement is like:
+
+    .. code:: text
+
+        (x0, y0) (x1, y1) (x2, y2)
+        (x3, y3) (x4, y4) (x5, y5)
+        (x6, y6) (x7, y7) (x8, y8)
+
+    Args:
+        in_channels (int): Same as nn.Conv2d.
+        out_channels (int): Same as nn.Conv2d.
+        kernel_size (int or tuple[int]): Same as nn.Conv2d.
+        stride (int or tuple[int]): Same as nn.Conv2d.
+        padding (int or tuple[int]): Same as nn.Conv2d.
+        dilation (int or tuple[int]): Same as nn.Conv2d.
+        groups (int): Same as nn.Conv2d.
+        bias (bool or str): If specified as `auto`, it will be decided by the
+            norm_cfg. Bias will be set as True if norm_cfg is None, otherwise
+            False.
+    """
+
+    _version = 2
+
+    def __init__(self, *args, **kwargs):
+        super(DeformConv2dPack, self).__init__(*args, **kwargs)
+        self.conv_offset = nn.Conv2d(
+            self.in_channels,
+            self.deform_groups * 2 * self.kernel_size[0] * self.kernel_size[1],
+            kernel_size=self.kernel_size,
+            stride=_pair(self.stride),
+            padding=_pair(self.padding),
+            dilation=_pair(self.dilation),
+            bias=True)
+        self.init_offset()
+
+    def init_offset(self):
+        self.conv_offset.weight.data.zero_()
+        self.conv_offset.bias.data.zero_()
+
+    def forward(self, x):
+        offset = self.conv_offset(x)
+        return deform_conv2d(x, offset, self.weight, self.stride, self.padding,
+                             self.dilation, self.groups, self.deform_groups,
+                             False, self.im2col_step)
+
+    def _load_from_state_dict(self, state_dict, prefix, local_metadata, strict,
+                              missing_keys, unexpected_keys, error_msgs):
+        version = local_metadata.get('version', None)
+
+        if version is None or version < 2:
+            # the key is different in early versions
+            # In version < 2, DeformConvPack loads previous benchmark models.
+            if (prefix + 'conv_offset.weight' not in state_dict
+                    and prefix[:-1] + '_offset.weight' in state_dict):
+                state_dict[prefix + 'conv_offset.weight'] = state_dict.pop(
+                    prefix[:-1] + '_offset.weight')
+            if (prefix + 'conv_offset.bias' not in state_dict
+                    and prefix[:-1] + '_offset.bias' in state_dict):
+                state_dict[prefix +
+                           'conv_offset.bias'] = state_dict.pop(prefix[:-1] +
+                                                                '_offset.bias')
+
+        if version is not None and version > 1:
+            print_log(
+                f'DeformConv2dPack {prefix.rstrip(".")} is upgraded to '
+                'version 2.',
+                logger='root')
+
+        super()._load_from_state_dict(state_dict, prefix, local_metadata,
+                                      strict, missing_keys, unexpected_keys,
+                                      error_msgs)
diff --git a/annotator/uniformer/mmcv/ops/deform_roi_pool.py b/annotator/uniformer/mmcv/ops/deform_roi_pool.py
new file mode 100644
index 0000000000000000000000000000000000000000..cc245ba91fee252226ba22e76bb94a35db9a629b
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/deform_roi_pool.py
@@ -0,0 +1,204 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from torch import nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.utils import _pair
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['deform_roi_pool_forward', 'deform_roi_pool_backward'])
+
+
+class DeformRoIPoolFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, rois, offset, output_size, spatial_scale,
+                 sampling_ratio, gamma):
+        return g.op(
+            'mmcv::MMCVDeformRoIPool',
+            input,
+            rois,
+            offset,
+            pooled_height_i=output_size[0],
+            pooled_width_i=output_size[1],
+            spatial_scale_f=spatial_scale,
+            sampling_ratio_f=sampling_ratio,
+            gamma_f=gamma)
+
+    @staticmethod
+    def forward(ctx,
+                input,
+                rois,
+                offset,
+                output_size,
+                spatial_scale=1.0,
+                sampling_ratio=0,
+                gamma=0.1):
+        if offset is None:
+            offset = input.new_zeros(0)
+        ctx.output_size = _pair(output_size)
+        ctx.spatial_scale = float(spatial_scale)
+        ctx.sampling_ratio = int(sampling_ratio)
+        ctx.gamma = float(gamma)
+
+        assert rois.size(1) == 5, 'RoI must be (idx, x1, y1, x2, y2)!'
+
+        output_shape = (rois.size(0), input.size(1), ctx.output_size[0],
+                        ctx.output_size[1])
+        output = input.new_zeros(output_shape)
+
+        ext_module.deform_roi_pool_forward(
+            input,
+            rois,
+            offset,
+            output,
+            pooled_height=ctx.output_size[0],
+            pooled_width=ctx.output_size[1],
+            spatial_scale=ctx.spatial_scale,
+            sampling_ratio=ctx.sampling_ratio,
+            gamma=ctx.gamma)
+
+        ctx.save_for_backward(input, rois, offset)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        input, rois, offset = ctx.saved_tensors
+        grad_input = grad_output.new_zeros(input.shape)
+        grad_offset = grad_output.new_zeros(offset.shape)
+
+        ext_module.deform_roi_pool_backward(
+            grad_output,
+            input,
+            rois,
+            offset,
+            grad_input,
+            grad_offset,
+            pooled_height=ctx.output_size[0],
+            pooled_width=ctx.output_size[1],
+            spatial_scale=ctx.spatial_scale,
+            sampling_ratio=ctx.sampling_ratio,
+            gamma=ctx.gamma)
+        if grad_offset.numel() == 0:
+            grad_offset = None
+        return grad_input, None, grad_offset, None, None, None, None
+
+
+deform_roi_pool = DeformRoIPoolFunction.apply
+
+
+class DeformRoIPool(nn.Module):
+
+    def __init__(self,
+                 output_size,
+                 spatial_scale=1.0,
+                 sampling_ratio=0,
+                 gamma=0.1):
+        super(DeformRoIPool, self).__init__()
+        self.output_size = _pair(output_size)
+        self.spatial_scale = float(spatial_scale)
+        self.sampling_ratio = int(sampling_ratio)
+        self.gamma = float(gamma)
+
+    def forward(self, input, rois, offset=None):
+        return deform_roi_pool(input, rois, offset, self.output_size,
+                               self.spatial_scale, self.sampling_ratio,
+                               self.gamma)
+
+
+class DeformRoIPoolPack(DeformRoIPool):
+
+    def __init__(self,
+                 output_size,
+                 output_channels,
+                 deform_fc_channels=1024,
+                 spatial_scale=1.0,
+                 sampling_ratio=0,
+                 gamma=0.1):
+        super(DeformRoIPoolPack, self).__init__(output_size, spatial_scale,
+                                                sampling_ratio, gamma)
+
+        self.output_channels = output_channels
+        self.deform_fc_channels = deform_fc_channels
+
+        self.offset_fc = nn.Sequential(
+            nn.Linear(
+                self.output_size[0] * self.output_size[1] *
+                self.output_channels, self.deform_fc_channels),
+            nn.ReLU(inplace=True),
+            nn.Linear(self.deform_fc_channels, self.deform_fc_channels),
+            nn.ReLU(inplace=True),
+            nn.Linear(self.deform_fc_channels,
+                      self.output_size[0] * self.output_size[1] * 2))
+        self.offset_fc[-1].weight.data.zero_()
+        self.offset_fc[-1].bias.data.zero_()
+
+    def forward(self, input, rois):
+        assert input.size(1) == self.output_channels
+        x = deform_roi_pool(input, rois, None, self.output_size,
+                            self.spatial_scale, self.sampling_ratio,
+                            self.gamma)
+        rois_num = rois.size(0)
+        offset = self.offset_fc(x.view(rois_num, -1))
+        offset = offset.view(rois_num, 2, self.output_size[0],
+                             self.output_size[1])
+        return deform_roi_pool(input, rois, offset, self.output_size,
+                               self.spatial_scale, self.sampling_ratio,
+                               self.gamma)
+
+
+class ModulatedDeformRoIPoolPack(DeformRoIPool):
+
+    def __init__(self,
+                 output_size,
+                 output_channels,
+                 deform_fc_channels=1024,
+                 spatial_scale=1.0,
+                 sampling_ratio=0,
+                 gamma=0.1):
+        super(ModulatedDeformRoIPoolPack,
+              self).__init__(output_size, spatial_scale, sampling_ratio, gamma)
+
+        self.output_channels = output_channels
+        self.deform_fc_channels = deform_fc_channels
+
+        self.offset_fc = nn.Sequential(
+            nn.Linear(
+                self.output_size[0] * self.output_size[1] *
+                self.output_channels, self.deform_fc_channels),
+            nn.ReLU(inplace=True),
+            nn.Linear(self.deform_fc_channels, self.deform_fc_channels),
+            nn.ReLU(inplace=True),
+            nn.Linear(self.deform_fc_channels,
+                      self.output_size[0] * self.output_size[1] * 2))
+        self.offset_fc[-1].weight.data.zero_()
+        self.offset_fc[-1].bias.data.zero_()
+
+        self.mask_fc = nn.Sequential(
+            nn.Linear(
+                self.output_size[0] * self.output_size[1] *
+                self.output_channels, self.deform_fc_channels),
+            nn.ReLU(inplace=True),
+            nn.Linear(self.deform_fc_channels,
+                      self.output_size[0] * self.output_size[1] * 1),
+            nn.Sigmoid())
+        self.mask_fc[2].weight.data.zero_()
+        self.mask_fc[2].bias.data.zero_()
+
+    def forward(self, input, rois):
+        assert input.size(1) == self.output_channels
+        x = deform_roi_pool(input, rois, None, self.output_size,
+                            self.spatial_scale, self.sampling_ratio,
+                            self.gamma)
+        rois_num = rois.size(0)
+        offset = self.offset_fc(x.view(rois_num, -1))
+        offset = offset.view(rois_num, 2, self.output_size[0],
+                             self.output_size[1])
+        mask = self.mask_fc(x.view(rois_num, -1))
+        mask = mask.view(rois_num, 1, self.output_size[0], self.output_size[1])
+        d = deform_roi_pool(input, rois, offset, self.output_size,
+                            self.spatial_scale, self.sampling_ratio,
+                            self.gamma)
+        return d * mask
diff --git a/annotator/uniformer/mmcv/ops/deprecated_wrappers.py b/annotator/uniformer/mmcv/ops/deprecated_wrappers.py
new file mode 100644
index 0000000000000000000000000000000000000000..a2e593df9ee57637038683d7a1efaa347b2b69e7
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/deprecated_wrappers.py
@@ -0,0 +1,43 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+# This file is for backward compatibility.
+# Module wrappers for empty tensor have been moved to mmcv.cnn.bricks.
+import warnings
+
+from ..cnn.bricks.wrappers import Conv2d, ConvTranspose2d, Linear, MaxPool2d
+
+
+class Conv2d_deprecated(Conv2d):
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        warnings.warn(
+            'Importing Conv2d wrapper from "mmcv.ops" will be deprecated in'
+            ' the future. Please import them from "mmcv.cnn" instead')
+
+
+class ConvTranspose2d_deprecated(ConvTranspose2d):
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        warnings.warn(
+            'Importing ConvTranspose2d wrapper from "mmcv.ops" will be '
+            'deprecated in the future. Please import them from "mmcv.cnn" '
+            'instead')
+
+
+class MaxPool2d_deprecated(MaxPool2d):
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        warnings.warn(
+            'Importing MaxPool2d wrapper from "mmcv.ops" will be deprecated in'
+            ' the future. Please import them from "mmcv.cnn" instead')
+
+
+class Linear_deprecated(Linear):
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        warnings.warn(
+            'Importing Linear wrapper from "mmcv.ops" will be deprecated in'
+            ' the future. Please import them from "mmcv.cnn" instead')
diff --git a/annotator/uniformer/mmcv/ops/focal_loss.py b/annotator/uniformer/mmcv/ops/focal_loss.py
new file mode 100644
index 0000000000000000000000000000000000000000..763bc93bd2575c49ca8ccf20996bbd92d1e0d1a4
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/focal_loss.py
@@ -0,0 +1,212 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', [
+    'sigmoid_focal_loss_forward', 'sigmoid_focal_loss_backward',
+    'softmax_focal_loss_forward', 'softmax_focal_loss_backward'
+])
+
+
+class SigmoidFocalLossFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, target, gamma, alpha, weight, reduction):
+        return g.op(
+            'mmcv::MMCVSigmoidFocalLoss',
+            input,
+            target,
+            gamma_f=gamma,
+            alpha_f=alpha,
+            weight_f=weight,
+            reduction_s=reduction)
+
+    @staticmethod
+    def forward(ctx,
+                input,
+                target,
+                gamma=2.0,
+                alpha=0.25,
+                weight=None,
+                reduction='mean'):
+
+        assert isinstance(target, (torch.LongTensor, torch.cuda.LongTensor))
+        assert input.dim() == 2
+        assert target.dim() == 1
+        assert input.size(0) == target.size(0)
+        if weight is None:
+            weight = input.new_empty(0)
+        else:
+            assert weight.dim() == 1
+            assert input.size(1) == weight.size(0)
+        ctx.reduction_dict = {'none': 0, 'mean': 1, 'sum': 2}
+        assert reduction in ctx.reduction_dict.keys()
+
+        ctx.gamma = float(gamma)
+        ctx.alpha = float(alpha)
+        ctx.reduction = ctx.reduction_dict[reduction]
+
+        output = input.new_zeros(input.size())
+
+        ext_module.sigmoid_focal_loss_forward(
+            input, target, weight, output, gamma=ctx.gamma, alpha=ctx.alpha)
+        if ctx.reduction == ctx.reduction_dict['mean']:
+            output = output.sum() / input.size(0)
+        elif ctx.reduction == ctx.reduction_dict['sum']:
+            output = output.sum()
+        ctx.save_for_backward(input, target, weight)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        input, target, weight = ctx.saved_tensors
+
+        grad_input = input.new_zeros(input.size())
+
+        ext_module.sigmoid_focal_loss_backward(
+            input,
+            target,
+            weight,
+            grad_input,
+            gamma=ctx.gamma,
+            alpha=ctx.alpha)
+
+        grad_input *= grad_output
+        if ctx.reduction == ctx.reduction_dict['mean']:
+            grad_input /= input.size(0)
+        return grad_input, None, None, None, None, None
+
+
+sigmoid_focal_loss = SigmoidFocalLossFunction.apply
+
+
+class SigmoidFocalLoss(nn.Module):
+
+    def __init__(self, gamma, alpha, weight=None, reduction='mean'):
+        super(SigmoidFocalLoss, self).__init__()
+        self.gamma = gamma
+        self.alpha = alpha
+        self.register_buffer('weight', weight)
+        self.reduction = reduction
+
+    def forward(self, input, target):
+        return sigmoid_focal_loss(input, target, self.gamma, self.alpha,
+                                  self.weight, self.reduction)
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(gamma={self.gamma}, '
+        s += f'alpha={self.alpha}, '
+        s += f'reduction={self.reduction})'
+        return s
+
+
+class SoftmaxFocalLossFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, target, gamma, alpha, weight, reduction):
+        return g.op(
+            'mmcv::MMCVSoftmaxFocalLoss',
+            input,
+            target,
+            gamma_f=gamma,
+            alpha_f=alpha,
+            weight_f=weight,
+            reduction_s=reduction)
+
+    @staticmethod
+    def forward(ctx,
+                input,
+                target,
+                gamma=2.0,
+                alpha=0.25,
+                weight=None,
+                reduction='mean'):
+
+        assert isinstance(target, (torch.LongTensor, torch.cuda.LongTensor))
+        assert input.dim() == 2
+        assert target.dim() == 1
+        assert input.size(0) == target.size(0)
+        if weight is None:
+            weight = input.new_empty(0)
+        else:
+            assert weight.dim() == 1
+            assert input.size(1) == weight.size(0)
+        ctx.reduction_dict = {'none': 0, 'mean': 1, 'sum': 2}
+        assert reduction in ctx.reduction_dict.keys()
+
+        ctx.gamma = float(gamma)
+        ctx.alpha = float(alpha)
+        ctx.reduction = ctx.reduction_dict[reduction]
+
+        channel_stats, _ = torch.max(input, dim=1)
+        input_softmax = input - channel_stats.unsqueeze(1).expand_as(input)
+        input_softmax.exp_()
+
+        channel_stats = input_softmax.sum(dim=1)
+        input_softmax /= channel_stats.unsqueeze(1).expand_as(input)
+
+        output = input.new_zeros(input.size(0))
+        ext_module.softmax_focal_loss_forward(
+            input_softmax,
+            target,
+            weight,
+            output,
+            gamma=ctx.gamma,
+            alpha=ctx.alpha)
+
+        if ctx.reduction == ctx.reduction_dict['mean']:
+            output = output.sum() / input.size(0)
+        elif ctx.reduction == ctx.reduction_dict['sum']:
+            output = output.sum()
+        ctx.save_for_backward(input_softmax, target, weight)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        input_softmax, target, weight = ctx.saved_tensors
+        buff = input_softmax.new_zeros(input_softmax.size(0))
+        grad_input = input_softmax.new_zeros(input_softmax.size())
+
+        ext_module.softmax_focal_loss_backward(
+            input_softmax,
+            target,
+            weight,
+            buff,
+            grad_input,
+            gamma=ctx.gamma,
+            alpha=ctx.alpha)
+
+        grad_input *= grad_output
+        if ctx.reduction == ctx.reduction_dict['mean']:
+            grad_input /= input_softmax.size(0)
+        return grad_input, None, None, None, None, None
+
+
+softmax_focal_loss = SoftmaxFocalLossFunction.apply
+
+
+class SoftmaxFocalLoss(nn.Module):
+
+    def __init__(self, gamma, alpha, weight=None, reduction='mean'):
+        super(SoftmaxFocalLoss, self).__init__()
+        self.gamma = gamma
+        self.alpha = alpha
+        self.register_buffer('weight', weight)
+        self.reduction = reduction
+
+    def forward(self, input, target):
+        return softmax_focal_loss(input, target, self.gamma, self.alpha,
+                                  self.weight, self.reduction)
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(gamma={self.gamma}, '
+        s += f'alpha={self.alpha}, '
+        s += f'reduction={self.reduction})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/furthest_point_sample.py b/annotator/uniformer/mmcv/ops/furthest_point_sample.py
new file mode 100644
index 0000000000000000000000000000000000000000..374b7a878f1972c183941af28ba1df216ac1a60f
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/furthest_point_sample.py
@@ -0,0 +1,83 @@
+import torch
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', [
+    'furthest_point_sampling_forward',
+    'furthest_point_sampling_with_dist_forward'
+])
+
+
+class FurthestPointSampling(Function):
+    """Uses iterative furthest point sampling to select a set of features whose
+    corresponding points have the furthest distance."""
+
+    @staticmethod
+    def forward(ctx, points_xyz: torch.Tensor,
+                num_points: int) -> torch.Tensor:
+        """
+        Args:
+            points_xyz (Tensor): (B, N, 3) where N > num_points.
+            num_points (int): Number of points in the sampled set.
+
+        Returns:
+             Tensor: (B, num_points) indices of the sampled points.
+        """
+        assert points_xyz.is_contiguous()
+
+        B, N = points_xyz.size()[:2]
+        output = torch.cuda.IntTensor(B, num_points)
+        temp = torch.cuda.FloatTensor(B, N).fill_(1e10)
+
+        ext_module.furthest_point_sampling_forward(
+            points_xyz,
+            temp,
+            output,
+            b=B,
+            n=N,
+            m=num_points,
+        )
+        if torch.__version__ != 'parrots':
+            ctx.mark_non_differentiable(output)
+        return output
+
+    @staticmethod
+    def backward(xyz, a=None):
+        return None, None
+
+
+class FurthestPointSamplingWithDist(Function):
+    """Uses iterative furthest point sampling to select a set of features whose
+    corresponding points have the furthest distance."""
+
+    @staticmethod
+    def forward(ctx, points_dist: torch.Tensor,
+                num_points: int) -> torch.Tensor:
+        """
+        Args:
+            points_dist (Tensor): (B, N, N) Distance between each point pair.
+            num_points (int): Number of points in the sampled set.
+
+        Returns:
+             Tensor: (B, num_points) indices of the sampled points.
+        """
+        assert points_dist.is_contiguous()
+
+        B, N, _ = points_dist.size()
+        output = points_dist.new_zeros([B, num_points], dtype=torch.int32)
+        temp = points_dist.new_zeros([B, N]).fill_(1e10)
+
+        ext_module.furthest_point_sampling_with_dist_forward(
+            points_dist, temp, output, b=B, n=N, m=num_points)
+        if torch.__version__ != 'parrots':
+            ctx.mark_non_differentiable(output)
+        return output
+
+    @staticmethod
+    def backward(xyz, a=None):
+        return None, None
+
+
+furthest_point_sample = FurthestPointSampling.apply
+furthest_point_sample_with_dist = FurthestPointSamplingWithDist.apply
diff --git a/annotator/uniformer/mmcv/ops/fused_bias_leakyrelu.py b/annotator/uniformer/mmcv/ops/fused_bias_leakyrelu.py
new file mode 100644
index 0000000000000000000000000000000000000000..6d12508469c6c8fa1884debece44c58d158cb6fa
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/fused_bias_leakyrelu.py
@@ -0,0 +1,268 @@
+# modified from https://github.com/rosinality/stylegan2-pytorch/blob/master/op/fused_act.py # noqa:E501
+
+# Copyright (c) 2021, NVIDIA Corporation. All rights reserved.
+# NVIDIA Source Code License for StyleGAN2 with Adaptive Discriminator
+# Augmentation (ADA)
+# =======================================================================
+
+# 1. Definitions
+
+# "Licensor" means any person or entity that distributes its Work.
+
+# "Software" means the original work of authorship made available under
+# this License.
+
+# "Work" means the Software and any additions to or derivative works of
+# the Software that are made available under this License.
+
+# The terms "reproduce," "reproduction," "derivative works," and
+# "distribution" have the meaning as provided under U.S. copyright law;
+# provided, however, that for the purposes of this License, derivative
+# works shall not include works that remain separable from, or merely
+# link (or bind by name) to the interfaces of, the Work.
+
+# Works, including the Software, are "made available" under this License
+# by including in or with the Work either (a) a copyright notice
+# referencing the applicability of this License to the Work, or (b) a
+# copy of this License.
+
+# 2. License Grants
+
+#     2.1 Copyright Grant. Subject to the terms and conditions of this
+#     License, each Licensor grants to you a perpetual, worldwide,
+#     non-exclusive, royalty-free, copyright license to reproduce,
+#     prepare derivative works of, publicly display, publicly perform,
+#     sublicense and distribute its Work and any resulting derivative
+#     works in any form.
+
+# 3. Limitations
+
+#     3.1 Redistribution. You may reproduce or distribute the Work only
+#     if (a) you do so under this License, (b) you include a complete
+#     copy of this License with your distribution, and (c) you retain
+#     without modification any copyright, patent, trademark, or
+#     attribution notices that are present in the Work.
+
+#     3.2 Derivative Works. You may specify that additional or different
+#     terms apply to the use, reproduction, and distribution of your
+#     derivative works of the Work ("Your Terms") only if (a) Your Terms
+#     provide that the use limitation in Section 3.3 applies to your
+#     derivative works, and (b) you identify the specific derivative
+#     works that are subject to Your Terms. Notwithstanding Your Terms,
+#     this License (including the redistribution requirements in Section
+#     3.1) will continue to apply to the Work itself.
+
+#     3.3 Use Limitation. The Work and any derivative works thereof only
+#     may be used or intended for use non-commercially. Notwithstanding
+#     the foregoing, NVIDIA and its affiliates may use the Work and any
+#     derivative works commercially. As used herein, "non-commercially"
+#     means for research or evaluation purposes only.
+
+#     3.4 Patent Claims. If you bring or threaten to bring a patent claim
+#     against any Licensor (including any claim, cross-claim or
+#     counterclaim in a lawsuit) to enforce any patents that you allege
+#     are infringed by any Work, then your rights under this License from
+#     such Licensor (including the grant in Section 2.1) will terminate
+#     immediately.
+
+#     3.5 Trademarks. This License does not grant any rights to use any
+#     Licensor’s or its affiliates’ names, logos, or trademarks, except
+#     as necessary to reproduce the notices described in this License.
+
+#     3.6 Termination. If you violate any term of this License, then your
+#     rights under this License (including the grant in Section 2.1) will
+#     terminate immediately.
+
+# 4. Disclaimer of Warranty.
+
+# THE WORK IS PROVIDED "AS IS" WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WARRANTIES OR CONDITIONS OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE OR
+# NON-INFRINGEMENT. YOU BEAR THE RISK OF UNDERTAKING ANY ACTIVITIES UNDER
+# THIS LICENSE.
+
+# 5. Limitation of Liability.
+
+# EXCEPT AS PROHIBITED BY APPLICABLE LAW, IN NO EVENT AND UNDER NO LEGAL
+# THEORY, WHETHER IN TORT (INCLUDING NEGLIGENCE), CONTRACT, OR OTHERWISE
+# SHALL ANY LICENSOR BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY DIRECT,
+# INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF
+# OR RELATED TO THIS LICENSE, THE USE OR INABILITY TO USE THE WORK
+# (INCLUDING BUT NOT LIMITED TO LOSS OF GOODWILL, BUSINESS INTERRUPTION,
+# LOST PROFITS OR DATA, COMPUTER FAILURE OR MALFUNCTION, OR ANY OTHER
+# COMMERCIAL DAMAGES OR LOSSES), EVEN IF THE LICENSOR HAS BEEN ADVISED OF
+# THE POSSIBILITY OF SUCH DAMAGES.
+
+# =======================================================================
+
+import torch
+import torch.nn.functional as F
+from torch import nn
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['fused_bias_leakyrelu'])
+
+
+class FusedBiasLeakyReLUFunctionBackward(Function):
+    """Calculate second order deviation.
+
+    This function is to compute the second order deviation for the fused leaky
+    relu operation.
+    """
+
+    @staticmethod
+    def forward(ctx, grad_output, out, negative_slope, scale):
+        ctx.save_for_backward(out)
+        ctx.negative_slope = negative_slope
+        ctx.scale = scale
+
+        empty = grad_output.new_empty(0)
+
+        grad_input = ext_module.fused_bias_leakyrelu(
+            grad_output,
+            empty,
+            out,
+            act=3,
+            grad=1,
+            alpha=negative_slope,
+            scale=scale)
+
+        dim = [0]
+
+        if grad_input.ndim > 2:
+            dim += list(range(2, grad_input.ndim))
+
+        grad_bias = grad_input.sum(dim).detach()
+
+        return grad_input, grad_bias
+
+    @staticmethod
+    def backward(ctx, gradgrad_input, gradgrad_bias):
+        out, = ctx.saved_tensors
+
+        # The second order deviation, in fact, contains two parts, while the
+        # the first part is zero. Thus, we direct consider the second part
+        # which is similar with the first order deviation in implementation.
+        gradgrad_out = ext_module.fused_bias_leakyrelu(
+            gradgrad_input,
+            gradgrad_bias.to(out.dtype),
+            out,
+            act=3,
+            grad=1,
+            alpha=ctx.negative_slope,
+            scale=ctx.scale)
+
+        return gradgrad_out, None, None, None
+
+
+class FusedBiasLeakyReLUFunction(Function):
+
+    @staticmethod
+    def forward(ctx, input, bias, negative_slope, scale):
+        empty = input.new_empty(0)
+
+        out = ext_module.fused_bias_leakyrelu(
+            input,
+            bias,
+            empty,
+            act=3,
+            grad=0,
+            alpha=negative_slope,
+            scale=scale)
+        ctx.save_for_backward(out)
+        ctx.negative_slope = negative_slope
+        ctx.scale = scale
+
+        return out
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        out, = ctx.saved_tensors
+
+        grad_input, grad_bias = FusedBiasLeakyReLUFunctionBackward.apply(
+            grad_output, out, ctx.negative_slope, ctx.scale)
+
+        return grad_input, grad_bias, None, None
+
+
+class FusedBiasLeakyReLU(nn.Module):
+    """Fused bias leaky ReLU.
+
+    This function is introduced in the StyleGAN2:
+    http://arxiv.org/abs/1912.04958
+
+    The bias term comes from the convolution operation. In addition, to keep
+    the variance of the feature map or gradients unchanged, they also adopt a
+    scale similarly with Kaiming initialization. However, since the
+    :math:`1+{alpha}^2` : is too small, we can just ignore it. Therefore, the
+    final scale is just :math:`\sqrt{2}`:. Of course, you may change it with # noqa: W605, E501
+    your own scale.
+
+    TODO: Implement the CPU version.
+
+    Args:
+        channel (int): The channel number of the feature map.
+        negative_slope (float, optional): Same as nn.LeakyRelu.
+            Defaults to 0.2.
+        scale (float, optional): A scalar to adjust the variance of the feature
+            map. Defaults to 2**0.5.
+    """
+
+    def __init__(self, num_channels, negative_slope=0.2, scale=2**0.5):
+        super(FusedBiasLeakyReLU, self).__init__()
+
+        self.bias = nn.Parameter(torch.zeros(num_channels))
+        self.negative_slope = negative_slope
+        self.scale = scale
+
+    def forward(self, input):
+        return fused_bias_leakyrelu(input, self.bias, self.negative_slope,
+                                    self.scale)
+
+
+def fused_bias_leakyrelu(input, bias, negative_slope=0.2, scale=2**0.5):
+    """Fused bias leaky ReLU function.
+
+    This function is introduced in the StyleGAN2:
+    http://arxiv.org/abs/1912.04958
+
+    The bias term comes from the convolution operation. In addition, to keep
+    the variance of the feature map or gradients unchanged, they also adopt a
+    scale similarly with Kaiming initialization. However, since the
+    :math:`1+{alpha}^2` : is too small, we can just ignore it. Therefore, the
+    final scale is just :math:`\sqrt{2}`:. Of course, you may change it with # noqa: W605, E501
+    your own scale.
+
+    Args:
+        input (torch.Tensor): Input feature map.
+        bias (nn.Parameter): The bias from convolution operation.
+        negative_slope (float, optional): Same as nn.LeakyRelu.
+            Defaults to 0.2.
+        scale (float, optional): A scalar to adjust the variance of the feature
+            map. Defaults to 2**0.5.
+
+    Returns:
+        torch.Tensor: Feature map after non-linear activation.
+    """
+
+    if not input.is_cuda:
+        return bias_leakyrelu_ref(input, bias, negative_slope, scale)
+
+    return FusedBiasLeakyReLUFunction.apply(input, bias.to(input.dtype),
+                                            negative_slope, scale)
+
+
+def bias_leakyrelu_ref(x, bias, negative_slope=0.2, scale=2**0.5):
+
+    if bias is not None:
+        assert bias.ndim == 1
+        assert bias.shape[0] == x.shape[1]
+        x = x + bias.reshape([-1 if i == 1 else 1 for i in range(x.ndim)])
+
+    x = F.leaky_relu(x, negative_slope)
+    if scale != 1:
+        x = x * scale
+
+    return x
diff --git a/annotator/uniformer/mmcv/ops/gather_points.py b/annotator/uniformer/mmcv/ops/gather_points.py
new file mode 100644
index 0000000000000000000000000000000000000000..f52f1677d8ea0facafc56a3672d37adb44677ff3
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/gather_points.py
@@ -0,0 +1,57 @@
+import torch
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['gather_points_forward', 'gather_points_backward'])
+
+
+class GatherPoints(Function):
+    """Gather points with given index."""
+
+    @staticmethod
+    def forward(ctx, features: torch.Tensor,
+                indices: torch.Tensor) -> torch.Tensor:
+        """
+        Args:
+            features (Tensor): (B, C, N) features to gather.
+            indices (Tensor): (B, M) where M is the number of points.
+
+        Returns:
+            Tensor: (B, C, M) where M is the number of points.
+        """
+        assert features.is_contiguous()
+        assert indices.is_contiguous()
+
+        B, npoint = indices.size()
+        _, C, N = features.size()
+        output = torch.cuda.FloatTensor(B, C, npoint)
+
+        ext_module.gather_points_forward(
+            features, indices, output, b=B, c=C, n=N, npoints=npoint)
+
+        ctx.for_backwards = (indices, C, N)
+        if torch.__version__ != 'parrots':
+            ctx.mark_non_differentiable(indices)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_out):
+        idx, C, N = ctx.for_backwards
+        B, npoint = idx.size()
+
+        grad_features = torch.cuda.FloatTensor(B, C, N).zero_()
+        grad_out_data = grad_out.data.contiguous()
+        ext_module.gather_points_backward(
+            grad_out_data,
+            idx,
+            grad_features.data,
+            b=B,
+            c=C,
+            n=N,
+            npoints=npoint)
+        return grad_features, None
+
+
+gather_points = GatherPoints.apply
diff --git a/annotator/uniformer/mmcv/ops/group_points.py b/annotator/uniformer/mmcv/ops/group_points.py
new file mode 100644
index 0000000000000000000000000000000000000000..6c3ec9d758ebe4e1c2205882af4be154008253a5
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/group_points.py
@@ -0,0 +1,224 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from typing import Tuple
+
+import torch
+from torch import nn as nn
+from torch.autograd import Function
+
+from ..utils import ext_loader
+from .ball_query import ball_query
+from .knn import knn
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['group_points_forward', 'group_points_backward'])
+
+
+class QueryAndGroup(nn.Module):
+    """Groups points with a ball query of radius.
+
+    Args:
+        max_radius (float): The maximum radius of the balls.
+            If None is given, we will use kNN sampling instead of ball query.
+        sample_num (int): Maximum number of features to gather in the ball.
+        min_radius (float, optional): The minimum radius of the balls.
+            Default: 0.
+        use_xyz (bool, optional): Whether to use xyz.
+            Default: True.
+        return_grouped_xyz (bool, optional): Whether to return grouped xyz.
+            Default: False.
+        normalize_xyz (bool, optional): Whether to normalize xyz.
+            Default: False.
+        uniform_sample (bool, optional): Whether to sample uniformly.
+            Default: False
+        return_unique_cnt (bool, optional): Whether to return the count of
+            unique samples. Default: False.
+        return_grouped_idx (bool, optional): Whether to return grouped idx.
+            Default: False.
+    """
+
+    def __init__(self,
+                 max_radius,
+                 sample_num,
+                 min_radius=0,
+                 use_xyz=True,
+                 return_grouped_xyz=False,
+                 normalize_xyz=False,
+                 uniform_sample=False,
+                 return_unique_cnt=False,
+                 return_grouped_idx=False):
+        super().__init__()
+        self.max_radius = max_radius
+        self.min_radius = min_radius
+        self.sample_num = sample_num
+        self.use_xyz = use_xyz
+        self.return_grouped_xyz = return_grouped_xyz
+        self.normalize_xyz = normalize_xyz
+        self.uniform_sample = uniform_sample
+        self.return_unique_cnt = return_unique_cnt
+        self.return_grouped_idx = return_grouped_idx
+        if self.return_unique_cnt:
+            assert self.uniform_sample, \
+                'uniform_sample should be True when ' \
+                'returning the count of unique samples'
+        if self.max_radius is None:
+            assert not self.normalize_xyz, \
+                'can not normalize grouped xyz when max_radius is None'
+
+    def forward(self, points_xyz, center_xyz, features=None):
+        """
+        Args:
+            points_xyz (Tensor): (B, N, 3) xyz coordinates of the features.
+            center_xyz (Tensor): (B, npoint, 3) coordinates of the centriods.
+            features (Tensor): (B, C, N) Descriptors of the features.
+
+        Returns:
+            Tensor: (B, 3 + C, npoint, sample_num) Grouped feature.
+        """
+        # if self.max_radius is None, we will perform kNN instead of ball query
+        # idx is of shape [B, npoint, sample_num]
+        if self.max_radius is None:
+            idx = knn(self.sample_num, points_xyz, center_xyz, False)
+            idx = idx.transpose(1, 2).contiguous()
+        else:
+            idx = ball_query(self.min_radius, self.max_radius, self.sample_num,
+                             points_xyz, center_xyz)
+
+        if self.uniform_sample:
+            unique_cnt = torch.zeros((idx.shape[0], idx.shape[1]))
+            for i_batch in range(idx.shape[0]):
+                for i_region in range(idx.shape[1]):
+                    unique_ind = torch.unique(idx[i_batch, i_region, :])
+                    num_unique = unique_ind.shape[0]
+                    unique_cnt[i_batch, i_region] = num_unique
+                    sample_ind = torch.randint(
+                        0,
+                        num_unique, (self.sample_num - num_unique, ),
+                        dtype=torch.long)
+                    all_ind = torch.cat((unique_ind, unique_ind[sample_ind]))
+                    idx[i_batch, i_region, :] = all_ind
+
+        xyz_trans = points_xyz.transpose(1, 2).contiguous()
+        # (B, 3, npoint, sample_num)
+        grouped_xyz = grouping_operation(xyz_trans, idx)
+        grouped_xyz_diff = grouped_xyz - \
+            center_xyz.transpose(1, 2).unsqueeze(-1)  # relative offsets
+        if self.normalize_xyz:
+            grouped_xyz_diff /= self.max_radius
+
+        if features is not None:
+            grouped_features = grouping_operation(features, idx)
+            if self.use_xyz:
+                # (B, C + 3, npoint, sample_num)
+                new_features = torch.cat([grouped_xyz_diff, grouped_features],
+                                         dim=1)
+            else:
+                new_features = grouped_features
+        else:
+            assert (self.use_xyz
+                    ), 'Cannot have not features and not use xyz as a feature!'
+            new_features = grouped_xyz_diff
+
+        ret = [new_features]
+        if self.return_grouped_xyz:
+            ret.append(grouped_xyz)
+        if self.return_unique_cnt:
+            ret.append(unique_cnt)
+        if self.return_grouped_idx:
+            ret.append(idx)
+        if len(ret) == 1:
+            return ret[0]
+        else:
+            return tuple(ret)
+
+
+class GroupAll(nn.Module):
+    """Group xyz with feature.
+
+    Args:
+        use_xyz (bool): Whether to use xyz.
+    """
+
+    def __init__(self, use_xyz: bool = True):
+        super().__init__()
+        self.use_xyz = use_xyz
+
+    def forward(self,
+                xyz: torch.Tensor,
+                new_xyz: torch.Tensor,
+                features: torch.Tensor = None):
+        """
+        Args:
+            xyz (Tensor): (B, N, 3) xyz coordinates of the features.
+            new_xyz (Tensor): new xyz coordinates of the features.
+            features (Tensor): (B, C, N) features to group.
+
+        Returns:
+            Tensor: (B, C + 3, 1, N) Grouped feature.
+        """
+        grouped_xyz = xyz.transpose(1, 2).unsqueeze(2)
+        if features is not None:
+            grouped_features = features.unsqueeze(2)
+            if self.use_xyz:
+                # (B, 3 + C, 1, N)
+                new_features = torch.cat([grouped_xyz, grouped_features],
+                                         dim=1)
+            else:
+                new_features = grouped_features
+        else:
+            new_features = grouped_xyz
+
+        return new_features
+
+
+class GroupingOperation(Function):
+    """Group feature with given index."""
+
+    @staticmethod
+    def forward(ctx, features: torch.Tensor,
+                indices: torch.Tensor) -> torch.Tensor:
+        """
+        Args:
+            features (Tensor): (B, C, N) tensor of features to group.
+            indices (Tensor): (B, npoint, nsample) the indices of
+                features to group with.
+
+        Returns:
+            Tensor: (B, C, npoint, nsample) Grouped features.
+        """
+        features = features.contiguous()
+        indices = indices.contiguous()
+
+        B, nfeatures, nsample = indices.size()
+        _, C, N = features.size()
+        output = torch.cuda.FloatTensor(B, C, nfeatures, nsample)
+
+        ext_module.group_points_forward(B, C, N, nfeatures, nsample, features,
+                                        indices, output)
+
+        ctx.for_backwards = (indices, N)
+        return output
+
+    @staticmethod
+    def backward(ctx,
+                 grad_out: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        """
+        Args:
+            grad_out (Tensor): (B, C, npoint, nsample) tensor of the gradients
+                of the output from forward.
+
+        Returns:
+            Tensor: (B, C, N) gradient of the features.
+        """
+        idx, N = ctx.for_backwards
+
+        B, C, npoint, nsample = grad_out.size()
+        grad_features = torch.cuda.FloatTensor(B, C, N).zero_()
+
+        grad_out_data = grad_out.data.contiguous()
+        ext_module.group_points_backward(B, C, N, npoint, nsample,
+                                         grad_out_data, idx,
+                                         grad_features.data)
+        return grad_features, None
+
+
+grouping_operation = GroupingOperation.apply
diff --git a/annotator/uniformer/mmcv/ops/info.py b/annotator/uniformer/mmcv/ops/info.py
new file mode 100644
index 0000000000000000000000000000000000000000..29f2e5598ae2bb5866ccd15a7d3b4de33c0cd14d
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/info.py
@@ -0,0 +1,36 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import glob
+import os
+
+import torch
+
+if torch.__version__ == 'parrots':
+    import parrots
+
+    def get_compiler_version():
+        return 'GCC ' + parrots.version.compiler
+
+    def get_compiling_cuda_version():
+        return parrots.version.cuda
+else:
+    from ..utils import ext_loader
+    ext_module = ext_loader.load_ext(
+        '_ext', ['get_compiler_version', 'get_compiling_cuda_version'])
+
+    def get_compiler_version():
+        return ext_module.get_compiler_version()
+
+    def get_compiling_cuda_version():
+        return ext_module.get_compiling_cuda_version()
+
+
+def get_onnxruntime_op_path():
+    wildcard = os.path.join(
+        os.path.abspath(os.path.dirname(os.path.dirname(__file__))),
+        '_ext_ort.*.so')
+
+    paths = glob.glob(wildcard)
+    if len(paths) > 0:
+        return paths[0]
+    else:
+        return ''
diff --git a/annotator/uniformer/mmcv/ops/iou3d.py b/annotator/uniformer/mmcv/ops/iou3d.py
new file mode 100644
index 0000000000000000000000000000000000000000..6fc71979190323f44c09f8b7e1761cf49cd2d76b
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/iou3d.py
@@ -0,0 +1,85 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', [
+    'iou3d_boxes_iou_bev_forward', 'iou3d_nms_forward',
+    'iou3d_nms_normal_forward'
+])
+
+
+def boxes_iou_bev(boxes_a, boxes_b):
+    """Calculate boxes IoU in the Bird's Eye View.
+
+    Args:
+        boxes_a (torch.Tensor): Input boxes a with shape (M, 5).
+        boxes_b (torch.Tensor): Input boxes b with shape (N, 5).
+
+    Returns:
+        ans_iou (torch.Tensor): IoU result with shape (M, N).
+    """
+    ans_iou = boxes_a.new_zeros(
+        torch.Size((boxes_a.shape[0], boxes_b.shape[0])))
+
+    ext_module.iou3d_boxes_iou_bev_forward(boxes_a.contiguous(),
+                                           boxes_b.contiguous(), ans_iou)
+
+    return ans_iou
+
+
+def nms_bev(boxes, scores, thresh, pre_max_size=None, post_max_size=None):
+    """NMS function GPU implementation (for BEV boxes). The overlap of two
+    boxes for IoU calculation is defined as the exact overlapping area of the
+    two boxes. In this function, one can also set ``pre_max_size`` and
+    ``post_max_size``.
+
+    Args:
+        boxes (torch.Tensor): Input boxes with the shape of [N, 5]
+            ([x1, y1, x2, y2, ry]).
+        scores (torch.Tensor): Scores of boxes with the shape of [N].
+        thresh (float): Overlap threshold of NMS.
+        pre_max_size (int, optional): Max size of boxes before NMS.
+            Default: None.
+        post_max_size (int, optional): Max size of boxes after NMS.
+            Default: None.
+
+    Returns:
+        torch.Tensor: Indexes after NMS.
+    """
+    assert boxes.size(1) == 5, 'Input boxes shape should be [N, 5]'
+    order = scores.sort(0, descending=True)[1]
+
+    if pre_max_size is not None:
+        order = order[:pre_max_size]
+    boxes = boxes[order].contiguous()
+
+    keep = torch.zeros(boxes.size(0), dtype=torch.long)
+    num_out = ext_module.iou3d_nms_forward(boxes, keep, thresh)
+    keep = order[keep[:num_out].cuda(boxes.device)].contiguous()
+    if post_max_size is not None:
+        keep = keep[:post_max_size]
+    return keep
+
+
+def nms_normal_bev(boxes, scores, thresh):
+    """Normal NMS function GPU implementation (for BEV boxes). The overlap of
+    two boxes for IoU calculation is defined as the exact overlapping area of
+    the two boxes WITH their yaw angle set to 0.
+
+    Args:
+        boxes (torch.Tensor): Input boxes with shape (N, 5).
+        scores (torch.Tensor): Scores of predicted boxes with shape (N).
+        thresh (float): Overlap threshold of NMS.
+
+    Returns:
+        torch.Tensor: Remaining indices with scores in descending order.
+    """
+    assert boxes.shape[1] == 5, 'Input boxes shape should be [N, 5]'
+    order = scores.sort(0, descending=True)[1]
+
+    boxes = boxes[order].contiguous()
+
+    keep = torch.zeros(boxes.size(0), dtype=torch.long)
+    num_out = ext_module.iou3d_nms_normal_forward(boxes, keep, thresh)
+    return order[keep[:num_out].cuda(boxes.device)].contiguous()
diff --git a/annotator/uniformer/mmcv/ops/knn.py b/annotator/uniformer/mmcv/ops/knn.py
new file mode 100644
index 0000000000000000000000000000000000000000..f335785036669fc19239825b0aae6dde3f73bf92
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/knn.py
@@ -0,0 +1,77 @@
+import torch
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['knn_forward'])
+
+
+class KNN(Function):
+    r"""KNN (CUDA) based on heap data structure.
+    Modified from `PAConv <https://github.com/CVMI-Lab/PAConv/tree/main/
+    scene_seg/lib/pointops/src/knnquery_heap>`_.
+
+    Find k-nearest points.
+    """
+
+    @staticmethod
+    def forward(ctx,
+                k: int,
+                xyz: torch.Tensor,
+                center_xyz: torch.Tensor = None,
+                transposed: bool = False) -> torch.Tensor:
+        """
+        Args:
+            k (int): number of nearest neighbors.
+            xyz (Tensor): (B, N, 3) if transposed == False, else (B, 3, N).
+                xyz coordinates of the features.
+            center_xyz (Tensor, optional): (B, npoint, 3) if transposed ==
+                False, else (B, 3, npoint). centers of the knn query.
+                Default: None.
+            transposed (bool, optional): whether the input tensors are
+                transposed. Should not explicitly use this keyword when
+                calling knn (=KNN.apply), just add the fourth param.
+                Default: False.
+
+        Returns:
+            Tensor: (B, k, npoint) tensor with the indices of
+                the features that form k-nearest neighbours.
+        """
+        assert (k > 0) & (k < 100), 'k should be in range(0, 100)'
+
+        if center_xyz is None:
+            center_xyz = xyz
+
+        if transposed:
+            xyz = xyz.transpose(2, 1).contiguous()
+            center_xyz = center_xyz.transpose(2, 1).contiguous()
+
+        assert xyz.is_contiguous()  # [B, N, 3]
+        assert center_xyz.is_contiguous()  # [B, npoint, 3]
+
+        center_xyz_device = center_xyz.get_device()
+        assert center_xyz_device == xyz.get_device(), \
+            'center_xyz and xyz should be put on the same device'
+        if torch.cuda.current_device() != center_xyz_device:
+            torch.cuda.set_device(center_xyz_device)
+
+        B, npoint, _ = center_xyz.shape
+        N = xyz.shape[1]
+
+        idx = center_xyz.new_zeros((B, npoint, k)).int()
+        dist2 = center_xyz.new_zeros((B, npoint, k)).float()
+
+        ext_module.knn_forward(
+            xyz, center_xyz, idx, dist2, b=B, n=N, m=npoint, nsample=k)
+        # idx shape to [B, k, npoint]
+        idx = idx.transpose(2, 1).contiguous()
+        if torch.__version__ != 'parrots':
+            ctx.mark_non_differentiable(idx)
+        return idx
+
+    @staticmethod
+    def backward(ctx, a=None):
+        return None, None, None
+
+
+knn = KNN.apply
diff --git a/annotator/uniformer/mmcv/ops/masked_conv.py b/annotator/uniformer/mmcv/ops/masked_conv.py
new file mode 100644
index 0000000000000000000000000000000000000000..cd514cc204c1d571ea5dc7e74b038c0f477a008b
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/masked_conv.py
@@ -0,0 +1,111 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import math
+
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.utils import _pair
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['masked_im2col_forward', 'masked_col2im_forward'])
+
+
+class MaskedConv2dFunction(Function):
+
+    @staticmethod
+    def symbolic(g, features, mask, weight, bias, padding, stride):
+        return g.op(
+            'mmcv::MMCVMaskedConv2d',
+            features,
+            mask,
+            weight,
+            bias,
+            padding_i=padding,
+            stride_i=stride)
+
+    @staticmethod
+    def forward(ctx, features, mask, weight, bias, padding=0, stride=1):
+        assert mask.dim() == 3 and mask.size(0) == 1
+        assert features.dim() == 4 and features.size(0) == 1
+        assert features.size()[2:] == mask.size()[1:]
+        pad_h, pad_w = _pair(padding)
+        stride_h, stride_w = _pair(stride)
+        if stride_h != 1 or stride_w != 1:
+            raise ValueError(
+                'Stride could not only be 1 in masked_conv2d currently.')
+        out_channel, in_channel, kernel_h, kernel_w = weight.size()
+
+        batch_size = features.size(0)
+        out_h = int(
+            math.floor((features.size(2) + 2 * pad_h -
+                        (kernel_h - 1) - 1) / stride_h + 1))
+        out_w = int(
+            math.floor((features.size(3) + 2 * pad_w -
+                        (kernel_h - 1) - 1) / stride_w + 1))
+        mask_inds = torch.nonzero(mask[0] > 0, as_tuple=False)
+        output = features.new_zeros(batch_size, out_channel, out_h, out_w)
+        if mask_inds.numel() > 0:
+            mask_h_idx = mask_inds[:, 0].contiguous()
+            mask_w_idx = mask_inds[:, 1].contiguous()
+            data_col = features.new_zeros(in_channel * kernel_h * kernel_w,
+                                          mask_inds.size(0))
+            ext_module.masked_im2col_forward(
+                features,
+                mask_h_idx,
+                mask_w_idx,
+                data_col,
+                kernel_h=kernel_h,
+                kernel_w=kernel_w,
+                pad_h=pad_h,
+                pad_w=pad_w)
+
+            masked_output = torch.addmm(1, bias[:, None], 1,
+                                        weight.view(out_channel, -1), data_col)
+            ext_module.masked_col2im_forward(
+                masked_output,
+                mask_h_idx,
+                mask_w_idx,
+                output,
+                height=out_h,
+                width=out_w,
+                channels=out_channel)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        return (None, ) * 5
+
+
+masked_conv2d = MaskedConv2dFunction.apply
+
+
+class MaskedConv2d(nn.Conv2d):
+    """A MaskedConv2d which inherits the official Conv2d.
+
+    The masked forward doesn't implement the backward function and only
+    supports the stride parameter to be 1 currently.
+    """
+
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 groups=1,
+                 bias=True):
+        super(MaskedConv2d,
+              self).__init__(in_channels, out_channels, kernel_size, stride,
+                             padding, dilation, groups, bias)
+
+    def forward(self, input, mask=None):
+        if mask is None:  # fallback to the normal Conv2d
+            return super(MaskedConv2d, self).forward(input)
+        else:
+            return masked_conv2d(input, mask, self.weight, self.bias,
+                                 self.padding)
diff --git a/annotator/uniformer/mmcv/ops/merge_cells.py b/annotator/uniformer/mmcv/ops/merge_cells.py
new file mode 100644
index 0000000000000000000000000000000000000000..48ca8cc0a8aca8432835bd760c0403a3c35b34cf
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/merge_cells.py
@@ -0,0 +1,149 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from abc import abstractmethod
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from ..cnn import ConvModule
+
+
+class BaseMergeCell(nn.Module):
+    """The basic class for cells used in NAS-FPN and NAS-FCOS.
+
+    BaseMergeCell takes 2 inputs. After applying convolution
+    on them, they are resized to the target size. Then,
+    they go through binary_op, which depends on the type of cell.
+    If with_out_conv is True, the result of output will go through
+    another convolution layer.
+
+    Args:
+        in_channels (int): number of input channels in out_conv layer.
+        out_channels (int): number of output channels in out_conv layer.
+        with_out_conv (bool): Whether to use out_conv layer
+        out_conv_cfg (dict): Config dict for convolution layer, which should
+            contain "groups", "kernel_size", "padding", "bias" to build
+            out_conv layer.
+        out_norm_cfg (dict): Config dict for normalization layer in out_conv.
+        out_conv_order (tuple): The order of conv/norm/activation layers in
+            out_conv.
+        with_input1_conv (bool): Whether to use convolution on input1.
+        with_input2_conv (bool): Whether to use convolution on input2.
+        input_conv_cfg (dict): Config dict for building input1_conv layer and
+            input2_conv layer, which is expected to contain the type of
+            convolution.
+            Default: None, which means using conv2d.
+        input_norm_cfg (dict): Config dict for normalization layer in
+            input1_conv and input2_conv layer. Default: None.
+        upsample_mode (str): Interpolation method used to resize the output
+            of input1_conv and input2_conv to target size. Currently, we
+            support ['nearest', 'bilinear']. Default: 'nearest'.
+    """
+
+    def __init__(self,
+                 fused_channels=256,
+                 out_channels=256,
+                 with_out_conv=True,
+                 out_conv_cfg=dict(
+                     groups=1, kernel_size=3, padding=1, bias=True),
+                 out_norm_cfg=None,
+                 out_conv_order=('act', 'conv', 'norm'),
+                 with_input1_conv=False,
+                 with_input2_conv=False,
+                 input_conv_cfg=None,
+                 input_norm_cfg=None,
+                 upsample_mode='nearest'):
+        super(BaseMergeCell, self).__init__()
+        assert upsample_mode in ['nearest', 'bilinear']
+        self.with_out_conv = with_out_conv
+        self.with_input1_conv = with_input1_conv
+        self.with_input2_conv = with_input2_conv
+        self.upsample_mode = upsample_mode
+
+        if self.with_out_conv:
+            self.out_conv = ConvModule(
+                fused_channels,
+                out_channels,
+                **out_conv_cfg,
+                norm_cfg=out_norm_cfg,
+                order=out_conv_order)
+
+        self.input1_conv = self._build_input_conv(
+            out_channels, input_conv_cfg,
+            input_norm_cfg) if with_input1_conv else nn.Sequential()
+        self.input2_conv = self._build_input_conv(
+            out_channels, input_conv_cfg,
+            input_norm_cfg) if with_input2_conv else nn.Sequential()
+
+    def _build_input_conv(self, channel, conv_cfg, norm_cfg):
+        return ConvModule(
+            channel,
+            channel,
+            3,
+            padding=1,
+            conv_cfg=conv_cfg,
+            norm_cfg=norm_cfg,
+            bias=True)
+
+    @abstractmethod
+    def _binary_op(self, x1, x2):
+        pass
+
+    def _resize(self, x, size):
+        if x.shape[-2:] == size:
+            return x
+        elif x.shape[-2:] < size:
+            return F.interpolate(x, size=size, mode=self.upsample_mode)
+        else:
+            assert x.shape[-2] % size[-2] == 0 and x.shape[-1] % size[-1] == 0
+            kernel_size = x.shape[-1] // size[-1]
+            x = F.max_pool2d(x, kernel_size=kernel_size, stride=kernel_size)
+            return x
+
+    def forward(self, x1, x2, out_size=None):
+        assert x1.shape[:2] == x2.shape[:2]
+        assert out_size is None or len(out_size) == 2
+        if out_size is None:  # resize to larger one
+            out_size = max(x1.size()[2:], x2.size()[2:])
+
+        x1 = self.input1_conv(x1)
+        x2 = self.input2_conv(x2)
+
+        x1 = self._resize(x1, out_size)
+        x2 = self._resize(x2, out_size)
+
+        x = self._binary_op(x1, x2)
+        if self.with_out_conv:
+            x = self.out_conv(x)
+        return x
+
+
+class SumCell(BaseMergeCell):
+
+    def __init__(self, in_channels, out_channels, **kwargs):
+        super(SumCell, self).__init__(in_channels, out_channels, **kwargs)
+
+    def _binary_op(self, x1, x2):
+        return x1 + x2
+
+
+class ConcatCell(BaseMergeCell):
+
+    def __init__(self, in_channels, out_channels, **kwargs):
+        super(ConcatCell, self).__init__(in_channels * 2, out_channels,
+                                         **kwargs)
+
+    def _binary_op(self, x1, x2):
+        ret = torch.cat([x1, x2], dim=1)
+        return ret
+
+
+class GlobalPoolingCell(BaseMergeCell):
+
+    def __init__(self, in_channels=None, out_channels=None, **kwargs):
+        super().__init__(in_channels, out_channels, **kwargs)
+        self.global_pool = nn.AdaptiveAvgPool2d((1, 1))
+
+    def _binary_op(self, x1, x2):
+        x2_att = self.global_pool(x2).sigmoid()
+        return x2 + x2_att * x1
diff --git a/annotator/uniformer/mmcv/ops/modulated_deform_conv.py b/annotator/uniformer/mmcv/ops/modulated_deform_conv.py
new file mode 100644
index 0000000000000000000000000000000000000000..75559579cf053abcc99538606cbb88c723faf783
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/modulated_deform_conv.py
@@ -0,0 +1,282 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import math
+
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.utils import _pair, _single
+
+from annotator.uniformer.mmcv.utils import deprecated_api_warning
+from ..cnn import CONV_LAYERS
+from ..utils import ext_loader, print_log
+
+ext_module = ext_loader.load_ext(
+    '_ext',
+    ['modulated_deform_conv_forward', 'modulated_deform_conv_backward'])
+
+
+class ModulatedDeformConv2dFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, offset, mask, weight, bias, stride, padding,
+                 dilation, groups, deform_groups):
+        input_tensors = [input, offset, mask, weight]
+        if bias is not None:
+            input_tensors.append(bias)
+        return g.op(
+            'mmcv::MMCVModulatedDeformConv2d',
+            *input_tensors,
+            stride_i=stride,
+            padding_i=padding,
+            dilation_i=dilation,
+            groups_i=groups,
+            deform_groups_i=deform_groups)
+
+    @staticmethod
+    def forward(ctx,
+                input,
+                offset,
+                mask,
+                weight,
+                bias=None,
+                stride=1,
+                padding=0,
+                dilation=1,
+                groups=1,
+                deform_groups=1):
+        if input is not None and input.dim() != 4:
+            raise ValueError(
+                f'Expected 4D tensor as input, got {input.dim()}D tensor \
+                  instead.')
+        ctx.stride = _pair(stride)
+        ctx.padding = _pair(padding)
+        ctx.dilation = _pair(dilation)
+        ctx.groups = groups
+        ctx.deform_groups = deform_groups
+        ctx.with_bias = bias is not None
+        if not ctx.with_bias:
+            bias = input.new_empty(0)  # fake tensor
+        # When pytorch version >= 1.6.0, amp is adopted for fp16 mode;
+        # amp won't cast the type of model (float32), but "offset" is cast
+        # to float16 by nn.Conv2d automatically, leading to the type
+        # mismatch with input (when it is float32) or weight.
+        # The flag for whether to use fp16 or amp is the type of "offset",
+        # we cast weight and input to temporarily support fp16 and amp
+        # whatever the pytorch version is.
+        input = input.type_as(offset)
+        weight = weight.type_as(input)
+        ctx.save_for_backward(input, offset, mask, weight, bias)
+        output = input.new_empty(
+            ModulatedDeformConv2dFunction._output_size(ctx, input, weight))
+        ctx._bufs = [input.new_empty(0), input.new_empty(0)]
+        ext_module.modulated_deform_conv_forward(
+            input,
+            weight,
+            bias,
+            ctx._bufs[0],
+            offset,
+            mask,
+            output,
+            ctx._bufs[1],
+            kernel_h=weight.size(2),
+            kernel_w=weight.size(3),
+            stride_h=ctx.stride[0],
+            stride_w=ctx.stride[1],
+            pad_h=ctx.padding[0],
+            pad_w=ctx.padding[1],
+            dilation_h=ctx.dilation[0],
+            dilation_w=ctx.dilation[1],
+            group=ctx.groups,
+            deformable_group=ctx.deform_groups,
+            with_bias=ctx.with_bias)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        input, offset, mask, weight, bias = ctx.saved_tensors
+        grad_input = torch.zeros_like(input)
+        grad_offset = torch.zeros_like(offset)
+        grad_mask = torch.zeros_like(mask)
+        grad_weight = torch.zeros_like(weight)
+        grad_bias = torch.zeros_like(bias)
+        grad_output = grad_output.contiguous()
+        ext_module.modulated_deform_conv_backward(
+            input,
+            weight,
+            bias,
+            ctx._bufs[0],
+            offset,
+            mask,
+            ctx._bufs[1],
+            grad_input,
+            grad_weight,
+            grad_bias,
+            grad_offset,
+            grad_mask,
+            grad_output,
+            kernel_h=weight.size(2),
+            kernel_w=weight.size(3),
+            stride_h=ctx.stride[0],
+            stride_w=ctx.stride[1],
+            pad_h=ctx.padding[0],
+            pad_w=ctx.padding[1],
+            dilation_h=ctx.dilation[0],
+            dilation_w=ctx.dilation[1],
+            group=ctx.groups,
+            deformable_group=ctx.deform_groups,
+            with_bias=ctx.with_bias)
+        if not ctx.with_bias:
+            grad_bias = None
+
+        return (grad_input, grad_offset, grad_mask, grad_weight, grad_bias,
+                None, None, None, None, None)
+
+    @staticmethod
+    def _output_size(ctx, input, weight):
+        channels = weight.size(0)
+        output_size = (input.size(0), channels)
+        for d in range(input.dim() - 2):
+            in_size = input.size(d + 2)
+            pad = ctx.padding[d]
+            kernel = ctx.dilation[d] * (weight.size(d + 2) - 1) + 1
+            stride_ = ctx.stride[d]
+            output_size += ((in_size + (2 * pad) - kernel) // stride_ + 1, )
+        if not all(map(lambda s: s > 0, output_size)):
+            raise ValueError(
+                'convolution input is too small (output would be ' +
+                'x'.join(map(str, output_size)) + ')')
+        return output_size
+
+
+modulated_deform_conv2d = ModulatedDeformConv2dFunction.apply
+
+
+class ModulatedDeformConv2d(nn.Module):
+
+    @deprecated_api_warning({'deformable_groups': 'deform_groups'},
+                            cls_name='ModulatedDeformConv2d')
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 groups=1,
+                 deform_groups=1,
+                 bias=True):
+        super(ModulatedDeformConv2d, self).__init__()
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.kernel_size = _pair(kernel_size)
+        self.stride = _pair(stride)
+        self.padding = _pair(padding)
+        self.dilation = _pair(dilation)
+        self.groups = groups
+        self.deform_groups = deform_groups
+        # enable compatibility with nn.Conv2d
+        self.transposed = False
+        self.output_padding = _single(0)
+
+        self.weight = nn.Parameter(
+            torch.Tensor(out_channels, in_channels // groups,
+                         *self.kernel_size))
+        if bias:
+            self.bias = nn.Parameter(torch.Tensor(out_channels))
+        else:
+            self.register_parameter('bias', None)
+        self.init_weights()
+
+    def init_weights(self):
+        n = self.in_channels
+        for k in self.kernel_size:
+            n *= k
+        stdv = 1. / math.sqrt(n)
+        self.weight.data.uniform_(-stdv, stdv)
+        if self.bias is not None:
+            self.bias.data.zero_()
+
+    def forward(self, x, offset, mask):
+        return modulated_deform_conv2d(x, offset, mask, self.weight, self.bias,
+                                       self.stride, self.padding,
+                                       self.dilation, self.groups,
+                                       self.deform_groups)
+
+
+@CONV_LAYERS.register_module('DCNv2')
+class ModulatedDeformConv2dPack(ModulatedDeformConv2d):
+    """A ModulatedDeformable Conv Encapsulation that acts as normal Conv
+    layers.
+
+    Args:
+        in_channels (int): Same as nn.Conv2d.
+        out_channels (int): Same as nn.Conv2d.
+        kernel_size (int or tuple[int]): Same as nn.Conv2d.
+        stride (int): Same as nn.Conv2d, while tuple is not supported.
+        padding (int): Same as nn.Conv2d, while tuple is not supported.
+        dilation (int): Same as nn.Conv2d, while tuple is not supported.
+        groups (int): Same as nn.Conv2d.
+        bias (bool or str): If specified as `auto`, it will be decided by the
+            norm_cfg. Bias will be set as True if norm_cfg is None, otherwise
+            False.
+    """
+
+    _version = 2
+
+    def __init__(self, *args, **kwargs):
+        super(ModulatedDeformConv2dPack, self).__init__(*args, **kwargs)
+        self.conv_offset = nn.Conv2d(
+            self.in_channels,
+            self.deform_groups * 3 * self.kernel_size[0] * self.kernel_size[1],
+            kernel_size=self.kernel_size,
+            stride=self.stride,
+            padding=self.padding,
+            dilation=self.dilation,
+            bias=True)
+        self.init_weights()
+
+    def init_weights(self):
+        super(ModulatedDeformConv2dPack, self).init_weights()
+        if hasattr(self, 'conv_offset'):
+            self.conv_offset.weight.data.zero_()
+            self.conv_offset.bias.data.zero_()
+
+    def forward(self, x):
+        out = self.conv_offset(x)
+        o1, o2, mask = torch.chunk(out, 3, dim=1)
+        offset = torch.cat((o1, o2), dim=1)
+        mask = torch.sigmoid(mask)
+        return modulated_deform_conv2d(x, offset, mask, self.weight, self.bias,
+                                       self.stride, self.padding,
+                                       self.dilation, self.groups,
+                                       self.deform_groups)
+
+    def _load_from_state_dict(self, state_dict, prefix, local_metadata, strict,
+                              missing_keys, unexpected_keys, error_msgs):
+        version = local_metadata.get('version', None)
+
+        if version is None or version < 2:
+            # the key is different in early versions
+            # In version < 2, ModulatedDeformConvPack
+            # loads previous benchmark models.
+            if (prefix + 'conv_offset.weight' not in state_dict
+                    and prefix[:-1] + '_offset.weight' in state_dict):
+                state_dict[prefix + 'conv_offset.weight'] = state_dict.pop(
+                    prefix[:-1] + '_offset.weight')
+            if (prefix + 'conv_offset.bias' not in state_dict
+                    and prefix[:-1] + '_offset.bias' in state_dict):
+                state_dict[prefix +
+                           'conv_offset.bias'] = state_dict.pop(prefix[:-1] +
+                                                                '_offset.bias')
+
+        if version is not None and version > 1:
+            print_log(
+                f'ModulatedDeformConvPack {prefix.rstrip(".")} is upgraded to '
+                'version 2.',
+                logger='root')
+
+        super()._load_from_state_dict(state_dict, prefix, local_metadata,
+                                      strict, missing_keys, unexpected_keys,
+                                      error_msgs)
diff --git a/annotator/uniformer/mmcv/ops/multi_scale_deform_attn.py b/annotator/uniformer/mmcv/ops/multi_scale_deform_attn.py
new file mode 100644
index 0000000000000000000000000000000000000000..c52dda18b41705705b47dd0e995b124048c16fba
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/multi_scale_deform_attn.py
@@ -0,0 +1,358 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import math
+import warnings
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.autograd.function import Function, once_differentiable
+
+from annotator.uniformer.mmcv import deprecated_api_warning
+from annotator.uniformer.mmcv.cnn import constant_init, xavier_init
+from annotator.uniformer.mmcv.cnn.bricks.registry import ATTENTION
+from annotator.uniformer.mmcv.runner import BaseModule
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['ms_deform_attn_backward', 'ms_deform_attn_forward'])
+
+
+class MultiScaleDeformableAttnFunction(Function):
+
+    @staticmethod
+    def forward(ctx, value, value_spatial_shapes, value_level_start_index,
+                sampling_locations, attention_weights, im2col_step):
+        """GPU version of multi-scale deformable attention.
+
+        Args:
+            value (Tensor): The value has shape
+                (bs, num_keys, mum_heads, embed_dims//num_heads)
+            value_spatial_shapes (Tensor): Spatial shape of
+                each feature map, has shape (num_levels, 2),
+                last dimension 2 represent (h, w)
+            sampling_locations (Tensor): The location of sampling points,
+                has shape
+                (bs ,num_queries, num_heads, num_levels, num_points, 2),
+                the last dimension 2 represent (x, y).
+            attention_weights (Tensor): The weight of sampling points used
+                when calculate the attention, has shape
+                (bs ,num_queries, num_heads, num_levels, num_points),
+            im2col_step (Tensor): The step used in image to column.
+
+        Returns:
+            Tensor: has shape (bs, num_queries, embed_dims)
+        """
+
+        ctx.im2col_step = im2col_step
+        output = ext_module.ms_deform_attn_forward(
+            value,
+            value_spatial_shapes,
+            value_level_start_index,
+            sampling_locations,
+            attention_weights,
+            im2col_step=ctx.im2col_step)
+        ctx.save_for_backward(value, value_spatial_shapes,
+                              value_level_start_index, sampling_locations,
+                              attention_weights)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        """GPU version of backward function.
+
+        Args:
+            grad_output (Tensor): Gradient
+                of output tensor of forward.
+
+        Returns:
+             Tuple[Tensor]: Gradient
+                of input tensors in forward.
+        """
+        value, value_spatial_shapes, value_level_start_index,\
+            sampling_locations, attention_weights = ctx.saved_tensors
+        grad_value = torch.zeros_like(value)
+        grad_sampling_loc = torch.zeros_like(sampling_locations)
+        grad_attn_weight = torch.zeros_like(attention_weights)
+
+        ext_module.ms_deform_attn_backward(
+            value,
+            value_spatial_shapes,
+            value_level_start_index,
+            sampling_locations,
+            attention_weights,
+            grad_output.contiguous(),
+            grad_value,
+            grad_sampling_loc,
+            grad_attn_weight,
+            im2col_step=ctx.im2col_step)
+
+        return grad_value, None, None, \
+            grad_sampling_loc, grad_attn_weight, None
+
+
+def multi_scale_deformable_attn_pytorch(value, value_spatial_shapes,
+                                        sampling_locations, attention_weights):
+    """CPU version of multi-scale deformable attention.
+
+    Args:
+        value (Tensor): The value has shape
+            (bs, num_keys, mum_heads, embed_dims//num_heads)
+        value_spatial_shapes (Tensor): Spatial shape of
+            each feature map, has shape (num_levels, 2),
+            last dimension 2 represent (h, w)
+        sampling_locations (Tensor): The location of sampling points,
+            has shape
+            (bs ,num_queries, num_heads, num_levels, num_points, 2),
+            the last dimension 2 represent (x, y).
+        attention_weights (Tensor): The weight of sampling points used
+            when calculate the attention, has shape
+            (bs ,num_queries, num_heads, num_levels, num_points),
+
+    Returns:
+        Tensor: has shape (bs, num_queries, embed_dims)
+    """
+
+    bs, _, num_heads, embed_dims = value.shape
+    _, num_queries, num_heads, num_levels, num_points, _ =\
+        sampling_locations.shape
+    value_list = value.split([H_ * W_ for H_, W_ in value_spatial_shapes],
+                             dim=1)
+    sampling_grids = 2 * sampling_locations - 1
+    sampling_value_list = []
+    for level, (H_, W_) in enumerate(value_spatial_shapes):
+        # bs, H_*W_, num_heads, embed_dims ->
+        # bs, H_*W_, num_heads*embed_dims ->
+        # bs, num_heads*embed_dims, H_*W_ ->
+        # bs*num_heads, embed_dims, H_, W_
+        value_l_ = value_list[level].flatten(2).transpose(1, 2).reshape(
+            bs * num_heads, embed_dims, H_, W_)
+        # bs, num_queries, num_heads, num_points, 2 ->
+        # bs, num_heads, num_queries, num_points, 2 ->
+        # bs*num_heads, num_queries, num_points, 2
+        sampling_grid_l_ = sampling_grids[:, :, :,
+                                          level].transpose(1, 2).flatten(0, 1)
+        # bs*num_heads, embed_dims, num_queries, num_points
+        sampling_value_l_ = F.grid_sample(
+            value_l_,
+            sampling_grid_l_,
+            mode='bilinear',
+            padding_mode='zeros',
+            align_corners=False)
+        sampling_value_list.append(sampling_value_l_)
+    # (bs, num_queries, num_heads, num_levels, num_points) ->
+    # (bs, num_heads, num_queries, num_levels, num_points) ->
+    # (bs, num_heads, 1, num_queries, num_levels*num_points)
+    attention_weights = attention_weights.transpose(1, 2).reshape(
+        bs * num_heads, 1, num_queries, num_levels * num_points)
+    output = (torch.stack(sampling_value_list, dim=-2).flatten(-2) *
+              attention_weights).sum(-1).view(bs, num_heads * embed_dims,
+                                              num_queries)
+    return output.transpose(1, 2).contiguous()
+
+
+@ATTENTION.register_module()
+class MultiScaleDeformableAttention(BaseModule):
+    """An attention module used in Deformable-Detr.
+
+    `Deformable DETR: Deformable Transformers for End-to-End Object Detection.
+    <https://arxiv.org/pdf/2010.04159.pdf>`_.
+
+    Args:
+        embed_dims (int): The embedding dimension of Attention.
+            Default: 256.
+        num_heads (int): Parallel attention heads. Default: 64.
+        num_levels (int): The number of feature map used in
+            Attention. Default: 4.
+        num_points (int): The number of sampling points for
+            each query in each head. Default: 4.
+        im2col_step (int): The step used in image_to_column.
+            Default: 64.
+        dropout (float): A Dropout layer on `inp_identity`.
+            Default: 0.1.
+        batch_first (bool): Key, Query and Value are shape of
+            (batch, n, embed_dim)
+            or (n, batch, embed_dim). Default to False.
+        norm_cfg (dict): Config dict for normalization layer.
+            Default: None.
+        init_cfg (obj:`mmcv.ConfigDict`): The Config for initialization.
+            Default: None.
+    """
+
+    def __init__(self,
+                 embed_dims=256,
+                 num_heads=8,
+                 num_levels=4,
+                 num_points=4,
+                 im2col_step=64,
+                 dropout=0.1,
+                 batch_first=False,
+                 norm_cfg=None,
+                 init_cfg=None):
+        super().__init__(init_cfg)
+        if embed_dims % num_heads != 0:
+            raise ValueError(f'embed_dims must be divisible by num_heads, '
+                             f'but got {embed_dims} and {num_heads}')
+        dim_per_head = embed_dims // num_heads
+        self.norm_cfg = norm_cfg
+        self.dropout = nn.Dropout(dropout)
+        self.batch_first = batch_first
+
+        # you'd better set dim_per_head to a power of 2
+        # which is more efficient in the CUDA implementation
+        def _is_power_of_2(n):
+            if (not isinstance(n, int)) or (n < 0):
+                raise ValueError(
+                    'invalid input for _is_power_of_2: {} (type: {})'.format(
+                        n, type(n)))
+            return (n & (n - 1) == 0) and n != 0
+
+        if not _is_power_of_2(dim_per_head):
+            warnings.warn(
+                "You'd better set embed_dims in "
+                'MultiScaleDeformAttention to make '
+                'the dimension of each attention head a power of 2 '
+                'which is more efficient in our CUDA implementation.')
+
+        self.im2col_step = im2col_step
+        self.embed_dims = embed_dims
+        self.num_levels = num_levels
+        self.num_heads = num_heads
+        self.num_points = num_points
+        self.sampling_offsets = nn.Linear(
+            embed_dims, num_heads * num_levels * num_points * 2)
+        self.attention_weights = nn.Linear(embed_dims,
+                                           num_heads * num_levels * num_points)
+        self.value_proj = nn.Linear(embed_dims, embed_dims)
+        self.output_proj = nn.Linear(embed_dims, embed_dims)
+        self.init_weights()
+
+    def init_weights(self):
+        """Default initialization for Parameters of Module."""
+        constant_init(self.sampling_offsets, 0.)
+        thetas = torch.arange(
+            self.num_heads,
+            dtype=torch.float32) * (2.0 * math.pi / self.num_heads)
+        grid_init = torch.stack([thetas.cos(), thetas.sin()], -1)
+        grid_init = (grid_init /
+                     grid_init.abs().max(-1, keepdim=True)[0]).view(
+                         self.num_heads, 1, 1,
+                         2).repeat(1, self.num_levels, self.num_points, 1)
+        for i in range(self.num_points):
+            grid_init[:, :, i, :] *= i + 1
+
+        self.sampling_offsets.bias.data = grid_init.view(-1)
+        constant_init(self.attention_weights, val=0., bias=0.)
+        xavier_init(self.value_proj, distribution='uniform', bias=0.)
+        xavier_init(self.output_proj, distribution='uniform', bias=0.)
+        self._is_init = True
+
+    @deprecated_api_warning({'residual': 'identity'},
+                            cls_name='MultiScaleDeformableAttention')
+    def forward(self,
+                query,
+                key=None,
+                value=None,
+                identity=None,
+                query_pos=None,
+                key_padding_mask=None,
+                reference_points=None,
+                spatial_shapes=None,
+                level_start_index=None,
+                **kwargs):
+        """Forward Function of MultiScaleDeformAttention.
+
+        Args:
+            query (Tensor): Query of Transformer with shape
+                (num_query, bs, embed_dims).
+            key (Tensor): The key tensor with shape
+                `(num_key, bs, embed_dims)`.
+            value (Tensor): The value tensor with shape
+                `(num_key, bs, embed_dims)`.
+            identity (Tensor): The tensor used for addition, with the
+                same shape as `query`. Default None. If None,
+                `query` will be used.
+            query_pos (Tensor): The positional encoding for `query`.
+                Default: None.
+            key_pos (Tensor): The positional encoding for `key`. Default
+                None.
+            reference_points (Tensor):  The normalized reference
+                points with shape (bs, num_query, num_levels, 2),
+                all elements is range in [0, 1], top-left (0,0),
+                bottom-right (1, 1), including padding area.
+                or (N, Length_{query}, num_levels, 4), add
+                additional two dimensions is (w, h) to
+                form reference boxes.
+            key_padding_mask (Tensor): ByteTensor for `query`, with
+                shape [bs, num_key].
+            spatial_shapes (Tensor): Spatial shape of features in
+                different levels. With shape (num_levels, 2),
+                last dimension represents (h, w).
+            level_start_index (Tensor): The start index of each level.
+                A tensor has shape ``(num_levels, )`` and can be represented
+                as [0, h_0*w_0, h_0*w_0+h_1*w_1, ...].
+
+        Returns:
+             Tensor: forwarded results with shape [num_query, bs, embed_dims].
+        """
+
+        if value is None:
+            value = query
+
+        if identity is None:
+            identity = query
+        if query_pos is not None:
+            query = query + query_pos
+        if not self.batch_first:
+            # change to (bs, num_query ,embed_dims)
+            query = query.permute(1, 0, 2)
+            value = value.permute(1, 0, 2)
+
+        bs, num_query, _ = query.shape
+        bs, num_value, _ = value.shape
+        assert (spatial_shapes[:, 0] * spatial_shapes[:, 1]).sum() == num_value
+
+        value = self.value_proj(value)
+        if key_padding_mask is not None:
+            value = value.masked_fill(key_padding_mask[..., None], 0.0)
+        value = value.view(bs, num_value, self.num_heads, -1)
+        sampling_offsets = self.sampling_offsets(query).view(
+            bs, num_query, self.num_heads, self.num_levels, self.num_points, 2)
+        attention_weights = self.attention_weights(query).view(
+            bs, num_query, self.num_heads, self.num_levels * self.num_points)
+        attention_weights = attention_weights.softmax(-1)
+
+        attention_weights = attention_weights.view(bs, num_query,
+                                                   self.num_heads,
+                                                   self.num_levels,
+                                                   self.num_points)
+        if reference_points.shape[-1] == 2:
+            offset_normalizer = torch.stack(
+                [spatial_shapes[..., 1], spatial_shapes[..., 0]], -1)
+            sampling_locations = reference_points[:, :, None, :, None, :] \
+                + sampling_offsets \
+                / offset_normalizer[None, None, None, :, None, :]
+        elif reference_points.shape[-1] == 4:
+            sampling_locations = reference_points[:, :, None, :, None, :2] \
+                + sampling_offsets / self.num_points \
+                * reference_points[:, :, None, :, None, 2:] \
+                * 0.5
+        else:
+            raise ValueError(
+                f'Last dim of reference_points must be'
+                f' 2 or 4, but get {reference_points.shape[-1]} instead.')
+        if torch.cuda.is_available() and value.is_cuda:
+            output = MultiScaleDeformableAttnFunction.apply(
+                value, spatial_shapes, level_start_index, sampling_locations,
+                attention_weights, self.im2col_step)
+        else:
+            output = multi_scale_deformable_attn_pytorch(
+                value, spatial_shapes, sampling_locations, attention_weights)
+
+        output = self.output_proj(output)
+
+        if not self.batch_first:
+            # (num_query, bs ,embed_dims)
+            output = output.permute(1, 0, 2)
+
+        return self.dropout(output) + identity
diff --git a/annotator/uniformer/mmcv/ops/nms.py b/annotator/uniformer/mmcv/ops/nms.py
new file mode 100644
index 0000000000000000000000000000000000000000..6d9634281f486ab284091786886854c451368052
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/nms.py
@@ -0,0 +1,417 @@
+import os
+
+import numpy as np
+import torch
+
+from annotator.uniformer.mmcv.utils import deprecated_api_warning
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['nms', 'softnms', 'nms_match', 'nms_rotated'])
+
+
+# This function is modified from: https://github.com/pytorch/vision/
+class NMSop(torch.autograd.Function):
+
+    @staticmethod
+    def forward(ctx, bboxes, scores, iou_threshold, offset, score_threshold,
+                max_num):
+        is_filtering_by_score = score_threshold > 0
+        if is_filtering_by_score:
+            valid_mask = scores > score_threshold
+            bboxes, scores = bboxes[valid_mask], scores[valid_mask]
+            valid_inds = torch.nonzero(
+                valid_mask, as_tuple=False).squeeze(dim=1)
+
+        inds = ext_module.nms(
+            bboxes, scores, iou_threshold=float(iou_threshold), offset=offset)
+
+        if max_num > 0:
+            inds = inds[:max_num]
+        if is_filtering_by_score:
+            inds = valid_inds[inds]
+        return inds
+
+    @staticmethod
+    def symbolic(g, bboxes, scores, iou_threshold, offset, score_threshold,
+                 max_num):
+        from ..onnx import is_custom_op_loaded
+        has_custom_op = is_custom_op_loaded()
+        # TensorRT nms plugin is aligned with original nms in ONNXRuntime
+        is_trt_backend = os.environ.get('ONNX_BACKEND') == 'MMCVTensorRT'
+        if has_custom_op and (not is_trt_backend):
+            return g.op(
+                'mmcv::NonMaxSuppression',
+                bboxes,
+                scores,
+                iou_threshold_f=float(iou_threshold),
+                offset_i=int(offset))
+        else:
+            from torch.onnx.symbolic_opset9 import select, squeeze, unsqueeze
+            from ..onnx.onnx_utils.symbolic_helper import _size_helper
+
+            boxes = unsqueeze(g, bboxes, 0)
+            scores = unsqueeze(g, unsqueeze(g, scores, 0), 0)
+
+            if max_num > 0:
+                max_num = g.op(
+                    'Constant',
+                    value_t=torch.tensor(max_num, dtype=torch.long))
+            else:
+                dim = g.op('Constant', value_t=torch.tensor(0))
+                max_num = _size_helper(g, bboxes, dim)
+            max_output_per_class = max_num
+            iou_threshold = g.op(
+                'Constant',
+                value_t=torch.tensor([iou_threshold], dtype=torch.float))
+            score_threshold = g.op(
+                'Constant',
+                value_t=torch.tensor([score_threshold], dtype=torch.float))
+            nms_out = g.op('NonMaxSuppression', boxes, scores,
+                           max_output_per_class, iou_threshold,
+                           score_threshold)
+            return squeeze(
+                g,
+                select(
+                    g, nms_out, 1,
+                    g.op(
+                        'Constant',
+                        value_t=torch.tensor([2], dtype=torch.long))), 1)
+
+
+class SoftNMSop(torch.autograd.Function):
+
+    @staticmethod
+    def forward(ctx, boxes, scores, iou_threshold, sigma, min_score, method,
+                offset):
+        dets = boxes.new_empty((boxes.size(0), 5), device='cpu')
+        inds = ext_module.softnms(
+            boxes.cpu(),
+            scores.cpu(),
+            dets.cpu(),
+            iou_threshold=float(iou_threshold),
+            sigma=float(sigma),
+            min_score=float(min_score),
+            method=int(method),
+            offset=int(offset))
+        return dets, inds
+
+    @staticmethod
+    def symbolic(g, boxes, scores, iou_threshold, sigma, min_score, method,
+                 offset):
+        from packaging import version
+        assert version.parse(torch.__version__) >= version.parse('1.7.0')
+        nms_out = g.op(
+            'mmcv::SoftNonMaxSuppression',
+            boxes,
+            scores,
+            iou_threshold_f=float(iou_threshold),
+            sigma_f=float(sigma),
+            min_score_f=float(min_score),
+            method_i=int(method),
+            offset_i=int(offset),
+            outputs=2)
+        return nms_out
+
+
+@deprecated_api_warning({'iou_thr': 'iou_threshold'})
+def nms(boxes, scores, iou_threshold, offset=0, score_threshold=0, max_num=-1):
+    """Dispatch to either CPU or GPU NMS implementations.
+
+    The input can be either torch tensor or numpy array. GPU NMS will be used
+    if the input is gpu tensor, otherwise CPU NMS
+    will be used. The returned type will always be the same as inputs.
+
+    Arguments:
+        boxes (torch.Tensor or np.ndarray): boxes in shape (N, 4).
+        scores (torch.Tensor or np.ndarray): scores in shape (N, ).
+        iou_threshold (float): IoU threshold for NMS.
+        offset (int, 0 or 1): boxes' width or height is (x2 - x1 + offset).
+        score_threshold (float): score threshold for NMS.
+        max_num (int): maximum number of boxes after NMS.
+
+    Returns:
+        tuple: kept dets(boxes and scores) and indice, which is always the \
+            same data type as the input.
+
+    Example:
+        >>> boxes = np.array([[49.1, 32.4, 51.0, 35.9],
+        >>>                   [49.3, 32.9, 51.0, 35.3],
+        >>>                   [49.2, 31.8, 51.0, 35.4],
+        >>>                   [35.1, 11.5, 39.1, 15.7],
+        >>>                   [35.6, 11.8, 39.3, 14.2],
+        >>>                   [35.3, 11.5, 39.9, 14.5],
+        >>>                   [35.2, 11.7, 39.7, 15.7]], dtype=np.float32)
+        >>> scores = np.array([0.9, 0.9, 0.5, 0.5, 0.5, 0.4, 0.3],\
+               dtype=np.float32)
+        >>> iou_threshold = 0.6
+        >>> dets, inds = nms(boxes, scores, iou_threshold)
+        >>> assert len(inds) == len(dets) == 3
+    """
+    assert isinstance(boxes, (torch.Tensor, np.ndarray))
+    assert isinstance(scores, (torch.Tensor, np.ndarray))
+    is_numpy = False
+    if isinstance(boxes, np.ndarray):
+        is_numpy = True
+        boxes = torch.from_numpy(boxes)
+    if isinstance(scores, np.ndarray):
+        scores = torch.from_numpy(scores)
+    assert boxes.size(1) == 4
+    assert boxes.size(0) == scores.size(0)
+    assert offset in (0, 1)
+
+    if torch.__version__ == 'parrots':
+        indata_list = [boxes, scores]
+        indata_dict = {
+            'iou_threshold': float(iou_threshold),
+            'offset': int(offset)
+        }
+        inds = ext_module.nms(*indata_list, **indata_dict)
+    else:
+        inds = NMSop.apply(boxes, scores, iou_threshold, offset,
+                           score_threshold, max_num)
+    dets = torch.cat((boxes[inds], scores[inds].reshape(-1, 1)), dim=1)
+    if is_numpy:
+        dets = dets.cpu().numpy()
+        inds = inds.cpu().numpy()
+    return dets, inds
+
+
+@deprecated_api_warning({'iou_thr': 'iou_threshold'})
+def soft_nms(boxes,
+             scores,
+             iou_threshold=0.3,
+             sigma=0.5,
+             min_score=1e-3,
+             method='linear',
+             offset=0):
+    """Dispatch to only CPU Soft NMS implementations.
+
+    The input can be either a torch tensor or numpy array.
+    The returned type will always be the same as inputs.
+
+    Arguments:
+        boxes (torch.Tensor or np.ndarray): boxes in shape (N, 4).
+        scores (torch.Tensor or np.ndarray): scores in shape (N, ).
+        iou_threshold (float): IoU threshold for NMS.
+        sigma (float): hyperparameter for gaussian method
+        min_score (float): score filter threshold
+        method (str): either 'linear' or 'gaussian'
+        offset (int, 0 or 1): boxes' width or height is (x2 - x1 + offset).
+
+    Returns:
+        tuple: kept dets(boxes and scores) and indice, which is always the \
+            same data type as the input.
+
+    Example:
+        >>> boxes = np.array([[4., 3., 5., 3.],
+        >>>                   [4., 3., 5., 4.],
+        >>>                   [3., 1., 3., 1.],
+        >>>                   [3., 1., 3., 1.],
+        >>>                   [3., 1., 3., 1.],
+        >>>                   [3., 1., 3., 1.]], dtype=np.float32)
+        >>> scores = np.array([0.9, 0.9, 0.5, 0.5, 0.4, 0.0], dtype=np.float32)
+        >>> iou_threshold = 0.6
+        >>> dets, inds = soft_nms(boxes, scores, iou_threshold, sigma=0.5)
+        >>> assert len(inds) == len(dets) == 5
+    """
+
+    assert isinstance(boxes, (torch.Tensor, np.ndarray))
+    assert isinstance(scores, (torch.Tensor, np.ndarray))
+    is_numpy = False
+    if isinstance(boxes, np.ndarray):
+        is_numpy = True
+        boxes = torch.from_numpy(boxes)
+    if isinstance(scores, np.ndarray):
+        scores = torch.from_numpy(scores)
+    assert boxes.size(1) == 4
+    assert boxes.size(0) == scores.size(0)
+    assert offset in (0, 1)
+    method_dict = {'naive': 0, 'linear': 1, 'gaussian': 2}
+    assert method in method_dict.keys()
+
+    if torch.__version__ == 'parrots':
+        dets = boxes.new_empty((boxes.size(0), 5), device='cpu')
+        indata_list = [boxes.cpu(), scores.cpu(), dets.cpu()]
+        indata_dict = {
+            'iou_threshold': float(iou_threshold),
+            'sigma': float(sigma),
+            'min_score': min_score,
+            'method': method_dict[method],
+            'offset': int(offset)
+        }
+        inds = ext_module.softnms(*indata_list, **indata_dict)
+    else:
+        dets, inds = SoftNMSop.apply(boxes.cpu(), scores.cpu(),
+                                     float(iou_threshold), float(sigma),
+                                     float(min_score), method_dict[method],
+                                     int(offset))
+
+    dets = dets[:inds.size(0)]
+
+    if is_numpy:
+        dets = dets.cpu().numpy()
+        inds = inds.cpu().numpy()
+        return dets, inds
+    else:
+        return dets.to(device=boxes.device), inds.to(device=boxes.device)
+
+
+def batched_nms(boxes, scores, idxs, nms_cfg, class_agnostic=False):
+    """Performs non-maximum suppression in a batched fashion.
+
+    Modified from https://github.com/pytorch/vision/blob
+    /505cd6957711af790211896d32b40291bea1bc21/torchvision/ops/boxes.py#L39.
+    In order to perform NMS independently per class, we add an offset to all
+    the boxes. The offset is dependent only on the class idx, and is large
+    enough so that boxes from different classes do not overlap.
+
+    Arguments:
+        boxes (torch.Tensor): boxes in shape (N, 4).
+        scores (torch.Tensor): scores in shape (N, ).
+        idxs (torch.Tensor): each index value correspond to a bbox cluster,
+            and NMS will not be applied between elements of different idxs,
+            shape (N, ).
+        nms_cfg (dict): specify nms type and other parameters like iou_thr.
+            Possible keys includes the following.
+
+            - iou_thr (float): IoU threshold used for NMS.
+            - split_thr (float): threshold number of boxes. In some cases the
+                number of boxes is large (e.g., 200k). To avoid OOM during
+                training, the users could set `split_thr` to a small value.
+                If the number of boxes is greater than the threshold, it will
+                perform NMS on each group of boxes separately and sequentially.
+                Defaults to 10000.
+        class_agnostic (bool): if true, nms is class agnostic,
+            i.e. IoU thresholding happens over all boxes,
+            regardless of the predicted class.
+
+    Returns:
+        tuple: kept dets and indice.
+    """
+    nms_cfg_ = nms_cfg.copy()
+    class_agnostic = nms_cfg_.pop('class_agnostic', class_agnostic)
+    if class_agnostic:
+        boxes_for_nms = boxes
+    else:
+        max_coordinate = boxes.max()
+        offsets = idxs.to(boxes) * (max_coordinate + torch.tensor(1).to(boxes))
+        boxes_for_nms = boxes + offsets[:, None]
+
+    nms_type = nms_cfg_.pop('type', 'nms')
+    nms_op = eval(nms_type)
+
+    split_thr = nms_cfg_.pop('split_thr', 10000)
+    # Won't split to multiple nms nodes when exporting to onnx
+    if boxes_for_nms.shape[0] < split_thr or torch.onnx.is_in_onnx_export():
+        dets, keep = nms_op(boxes_for_nms, scores, **nms_cfg_)
+        boxes = boxes[keep]
+        # -1 indexing works abnormal in TensorRT
+        # This assumes `dets` has 5 dimensions where
+        # the last dimension is score.
+        # TODO: more elegant way to handle the dimension issue.
+        # Some type of nms would reweight the score, such as SoftNMS
+        scores = dets[:, 4]
+    else:
+        max_num = nms_cfg_.pop('max_num', -1)
+        total_mask = scores.new_zeros(scores.size(), dtype=torch.bool)
+        # Some type of nms would reweight the score, such as SoftNMS
+        scores_after_nms = scores.new_zeros(scores.size())
+        for id in torch.unique(idxs):
+            mask = (idxs == id).nonzero(as_tuple=False).view(-1)
+            dets, keep = nms_op(boxes_for_nms[mask], scores[mask], **nms_cfg_)
+            total_mask[mask[keep]] = True
+            scores_after_nms[mask[keep]] = dets[:, -1]
+        keep = total_mask.nonzero(as_tuple=False).view(-1)
+
+        scores, inds = scores_after_nms[keep].sort(descending=True)
+        keep = keep[inds]
+        boxes = boxes[keep]
+
+        if max_num > 0:
+            keep = keep[:max_num]
+            boxes = boxes[:max_num]
+            scores = scores[:max_num]
+
+    return torch.cat([boxes, scores[:, None]], -1), keep
+
+
+def nms_match(dets, iou_threshold):
+    """Matched dets into different groups by NMS.
+
+    NMS match is Similar to NMS but when a bbox is suppressed, nms match will
+    record the indice of suppressed bbox and form a group with the indice of
+    kept bbox. In each group, indice is sorted as score order.
+
+    Arguments:
+        dets (torch.Tensor | np.ndarray): Det boxes with scores, shape (N, 5).
+        iou_thr (float): IoU thresh for NMS.
+
+    Returns:
+        List[torch.Tensor | np.ndarray]: The outer list corresponds different
+            matched group, the inner Tensor corresponds the indices for a group
+            in score order.
+    """
+    if dets.shape[0] == 0:
+        matched = []
+    else:
+        assert dets.shape[-1] == 5, 'inputs dets.shape should be (N, 5), ' \
+                                    f'but get {dets.shape}'
+        if isinstance(dets, torch.Tensor):
+            dets_t = dets.detach().cpu()
+        else:
+            dets_t = torch.from_numpy(dets)
+        indata_list = [dets_t]
+        indata_dict = {'iou_threshold': float(iou_threshold)}
+        matched = ext_module.nms_match(*indata_list, **indata_dict)
+        if torch.__version__ == 'parrots':
+            matched = matched.tolist()
+
+    if isinstance(dets, torch.Tensor):
+        return [dets.new_tensor(m, dtype=torch.long) for m in matched]
+    else:
+        return [np.array(m, dtype=np.int) for m in matched]
+
+
+def nms_rotated(dets, scores, iou_threshold, labels=None):
+    """Performs non-maximum suppression (NMS) on the rotated boxes according to
+    their intersection-over-union (IoU).
+
+    Rotated NMS iteratively removes lower scoring rotated boxes which have an
+    IoU greater than iou_threshold with another (higher scoring) rotated box.
+
+    Args:
+        boxes (Tensor):  Rotated boxes in shape (N, 5). They are expected to \
+            be in (x_ctr, y_ctr, width, height, angle_radian) format.
+        scores (Tensor): scores in shape (N, ).
+        iou_threshold (float): IoU thresh for NMS.
+        labels (Tensor): boxes' label in shape (N,).
+
+    Returns:
+        tuple: kept dets(boxes and scores) and indice, which is always the \
+            same data type as the input.
+    """
+    if dets.shape[0] == 0:
+        return dets, None
+    multi_label = labels is not None
+    if multi_label:
+        dets_wl = torch.cat((dets, labels.unsqueeze(1)), 1)
+    else:
+        dets_wl = dets
+    _, order = scores.sort(0, descending=True)
+    dets_sorted = dets_wl.index_select(0, order)
+
+    if torch.__version__ == 'parrots':
+        keep_inds = ext_module.nms_rotated(
+            dets_wl,
+            scores,
+            order,
+            dets_sorted,
+            iou_threshold=iou_threshold,
+            multi_label=multi_label)
+    else:
+        keep_inds = ext_module.nms_rotated(dets_wl, scores, order, dets_sorted,
+                                           iou_threshold, multi_label)
+    dets = torch.cat((dets[keep_inds], scores[keep_inds].reshape(-1, 1)),
+                     dim=1)
+    return dets, keep_inds
diff --git a/annotator/uniformer/mmcv/ops/pixel_group.py b/annotator/uniformer/mmcv/ops/pixel_group.py
new file mode 100644
index 0000000000000000000000000000000000000000..2143c75f835a467c802fc3c37ecd3ac0f85bcda4
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/pixel_group.py
@@ -0,0 +1,75 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import numpy as np
+import torch
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['pixel_group'])
+
+
+def pixel_group(score, mask, embedding, kernel_label, kernel_contour,
+                kernel_region_num, distance_threshold):
+    """Group pixels into text instances, which is widely used text detection
+    methods.
+
+    Arguments:
+        score (np.array or Tensor): The foreground score with size hxw.
+        mask (np.array or Tensor): The foreground mask with size hxw.
+        embedding (np.array or Tensor): The embedding with size hxwxc to
+            distinguish instances.
+        kernel_label (np.array or Tensor): The instance kernel index with
+            size hxw.
+        kernel_contour (np.array or Tensor): The kernel contour with size hxw.
+        kernel_region_num (int): The instance kernel region number.
+        distance_threshold (float): The embedding distance threshold between
+            kernel and pixel in one instance.
+
+    Returns:
+        pixel_assignment (List[List[float]]): The instance coordinate list.
+            Each element consists of averaged confidence, pixel number, and
+            coordinates (x_i, y_i for all pixels) in order.
+    """
+    assert isinstance(score, (torch.Tensor, np.ndarray))
+    assert isinstance(mask, (torch.Tensor, np.ndarray))
+    assert isinstance(embedding, (torch.Tensor, np.ndarray))
+    assert isinstance(kernel_label, (torch.Tensor, np.ndarray))
+    assert isinstance(kernel_contour, (torch.Tensor, np.ndarray))
+    assert isinstance(kernel_region_num, int)
+    assert isinstance(distance_threshold, float)
+
+    if isinstance(score, np.ndarray):
+        score = torch.from_numpy(score)
+    if isinstance(mask, np.ndarray):
+        mask = torch.from_numpy(mask)
+    if isinstance(embedding, np.ndarray):
+        embedding = torch.from_numpy(embedding)
+    if isinstance(kernel_label, np.ndarray):
+        kernel_label = torch.from_numpy(kernel_label)
+    if isinstance(kernel_contour, np.ndarray):
+        kernel_contour = torch.from_numpy(kernel_contour)
+
+    if torch.__version__ == 'parrots':
+        label = ext_module.pixel_group(
+            score,
+            mask,
+            embedding,
+            kernel_label,
+            kernel_contour,
+            kernel_region_num=kernel_region_num,
+            distance_threshold=distance_threshold)
+        label = label.tolist()
+        label = label[0]
+        list_index = kernel_region_num
+        pixel_assignment = []
+        for x in range(kernel_region_num):
+            pixel_assignment.append(
+                np.array(
+                    label[list_index:list_index + int(label[x])],
+                    dtype=np.float))
+            list_index = list_index + int(label[x])
+    else:
+        pixel_assignment = ext_module.pixel_group(score, mask, embedding,
+                                                  kernel_label, kernel_contour,
+                                                  kernel_region_num,
+                                                  distance_threshold)
+    return pixel_assignment
diff --git a/annotator/uniformer/mmcv/ops/point_sample.py b/annotator/uniformer/mmcv/ops/point_sample.py
new file mode 100644
index 0000000000000000000000000000000000000000..267f4b3c56630acd85f9bdc630b7be09abab0aba
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/point_sample.py
@@ -0,0 +1,336 @@
+# Modified from https://github.com/facebookresearch/detectron2/tree/master/projects/PointRend  # noqa
+
+from os import path as osp
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.nn.modules.utils import _pair
+from torch.onnx.operators import shape_as_tensor
+
+
+def bilinear_grid_sample(im, grid, align_corners=False):
+    """Given an input and a flow-field grid, computes the output using input
+    values and pixel locations from grid. Supported only bilinear interpolation
+    method to sample the input pixels.
+
+    Args:
+        im (torch.Tensor): Input feature map, shape (N, C, H, W)
+        grid (torch.Tensor): Point coordinates, shape (N, Hg, Wg, 2)
+        align_corners {bool}: If set to True, the extrema (-1 and 1) are
+            considered as referring to the center points of the input’s
+            corner pixels. If set to False, they are instead considered as
+            referring to the corner points of the input’s corner pixels,
+            making the sampling more resolution agnostic.
+    Returns:
+        torch.Tensor: A tensor with sampled points, shape (N, C, Hg, Wg)
+    """
+    n, c, h, w = im.shape
+    gn, gh, gw, _ = grid.shape
+    assert n == gn
+
+    x = grid[:, :, :, 0]
+    y = grid[:, :, :, 1]
+
+    if align_corners:
+        x = ((x + 1) / 2) * (w - 1)
+        y = ((y + 1) / 2) * (h - 1)
+    else:
+        x = ((x + 1) * w - 1) / 2
+        y = ((y + 1) * h - 1) / 2
+
+    x = x.view(n, -1)
+    y = y.view(n, -1)
+
+    x0 = torch.floor(x).long()
+    y0 = torch.floor(y).long()
+    x1 = x0 + 1
+    y1 = y0 + 1
+
+    wa = ((x1 - x) * (y1 - y)).unsqueeze(1)
+    wb = ((x1 - x) * (y - y0)).unsqueeze(1)
+    wc = ((x - x0) * (y1 - y)).unsqueeze(1)
+    wd = ((x - x0) * (y - y0)).unsqueeze(1)
+
+    # Apply default for grid_sample function zero padding
+    im_padded = F.pad(im, pad=[1, 1, 1, 1], mode='constant', value=0)
+    padded_h = h + 2
+    padded_w = w + 2
+    # save points positions after padding
+    x0, x1, y0, y1 = x0 + 1, x1 + 1, y0 + 1, y1 + 1
+
+    # Clip coordinates to padded image size
+    x0 = torch.where(x0 < 0, torch.tensor(0), x0)
+    x0 = torch.where(x0 > padded_w - 1, torch.tensor(padded_w - 1), x0)
+    x1 = torch.where(x1 < 0, torch.tensor(0), x1)
+    x1 = torch.where(x1 > padded_w - 1, torch.tensor(padded_w - 1), x1)
+    y0 = torch.where(y0 < 0, torch.tensor(0), y0)
+    y0 = torch.where(y0 > padded_h - 1, torch.tensor(padded_h - 1), y0)
+    y1 = torch.where(y1 < 0, torch.tensor(0), y1)
+    y1 = torch.where(y1 > padded_h - 1, torch.tensor(padded_h - 1), y1)
+
+    im_padded = im_padded.view(n, c, -1)
+
+    x0_y0 = (x0 + y0 * padded_w).unsqueeze(1).expand(-1, c, -1)
+    x0_y1 = (x0 + y1 * padded_w).unsqueeze(1).expand(-1, c, -1)
+    x1_y0 = (x1 + y0 * padded_w).unsqueeze(1).expand(-1, c, -1)
+    x1_y1 = (x1 + y1 * padded_w).unsqueeze(1).expand(-1, c, -1)
+
+    Ia = torch.gather(im_padded, 2, x0_y0)
+    Ib = torch.gather(im_padded, 2, x0_y1)
+    Ic = torch.gather(im_padded, 2, x1_y0)
+    Id = torch.gather(im_padded, 2, x1_y1)
+
+    return (Ia * wa + Ib * wb + Ic * wc + Id * wd).reshape(n, c, gh, gw)
+
+
+def is_in_onnx_export_without_custom_ops():
+    from annotator.uniformer.mmcv.ops import get_onnxruntime_op_path
+    ort_custom_op_path = get_onnxruntime_op_path()
+    return torch.onnx.is_in_onnx_export(
+    ) and not osp.exists(ort_custom_op_path)
+
+
+def normalize(grid):
+    """Normalize input grid from [-1, 1] to [0, 1]
+    Args:
+        grid (Tensor): The grid to be normalize, range [-1, 1].
+    Returns:
+        Tensor: Normalized grid, range [0, 1].
+    """
+
+    return (grid + 1.0) / 2.0
+
+
+def denormalize(grid):
+    """Denormalize input grid from range [0, 1] to [-1, 1]
+    Args:
+        grid (Tensor): The grid to be denormalize, range [0, 1].
+    Returns:
+        Tensor: Denormalized grid, range [-1, 1].
+    """
+
+    return grid * 2.0 - 1.0
+
+
+def generate_grid(num_grid, size, device):
+    """Generate regular square grid of points in [0, 1] x [0, 1] coordinate
+    space.
+
+    Args:
+        num_grid (int): The number of grids to sample, one for each region.
+        size (tuple(int, int)): The side size of the regular grid.
+        device (torch.device): Desired device of returned tensor.
+
+    Returns:
+        (torch.Tensor): A tensor of shape (num_grid, size[0]*size[1], 2) that
+            contains coordinates for the regular grids.
+    """
+
+    affine_trans = torch.tensor([[[1., 0., 0.], [0., 1., 0.]]], device=device)
+    grid = F.affine_grid(
+        affine_trans, torch.Size((1, 1, *size)), align_corners=False)
+    grid = normalize(grid)
+    return grid.view(1, -1, 2).expand(num_grid, -1, -1)
+
+
+def rel_roi_point_to_abs_img_point(rois, rel_roi_points):
+    """Convert roi based relative point coordinates to image based absolute
+    point coordinates.
+
+    Args:
+        rois (Tensor): RoIs or BBoxes, shape (N, 4) or (N, 5)
+        rel_roi_points (Tensor): Point coordinates inside RoI, relative to
+            RoI, location, range (0, 1), shape (N, P, 2)
+    Returns:
+        Tensor: Image based absolute point coordinates, shape (N, P, 2)
+    """
+
+    with torch.no_grad():
+        assert rel_roi_points.size(0) == rois.size(0)
+        assert rois.dim() == 2
+        assert rel_roi_points.dim() == 3
+        assert rel_roi_points.size(2) == 2
+        # remove batch idx
+        if rois.size(1) == 5:
+            rois = rois[:, 1:]
+        abs_img_points = rel_roi_points.clone()
+        # To avoid an error during exporting to onnx use independent
+        # variables instead inplace computation
+        xs = abs_img_points[:, :, 0] * (rois[:, None, 2] - rois[:, None, 0])
+        ys = abs_img_points[:, :, 1] * (rois[:, None, 3] - rois[:, None, 1])
+        xs += rois[:, None, 0]
+        ys += rois[:, None, 1]
+        abs_img_points = torch.stack([xs, ys], dim=2)
+    return abs_img_points
+
+
+def get_shape_from_feature_map(x):
+    """Get spatial resolution of input feature map considering exporting to
+    onnx mode.
+
+    Args:
+        x (torch.Tensor): Input tensor, shape (N, C, H, W)
+    Returns:
+        torch.Tensor: Spatial resolution (width, height), shape (1, 1, 2)
+    """
+    if torch.onnx.is_in_onnx_export():
+        img_shape = shape_as_tensor(x)[2:].flip(0).view(1, 1, 2).to(
+            x.device).float()
+    else:
+        img_shape = torch.tensor(x.shape[2:]).flip(0).view(1, 1, 2).to(
+            x.device).float()
+    return img_shape
+
+
+def abs_img_point_to_rel_img_point(abs_img_points, img, spatial_scale=1.):
+    """Convert image based absolute point coordinates to image based relative
+    coordinates for sampling.
+
+    Args:
+        abs_img_points (Tensor): Image based absolute point coordinates,
+            shape (N, P, 2)
+        img (tuple/Tensor): (height, width) of image or feature map.
+        spatial_scale (float): Scale points by this factor. Default: 1.
+
+    Returns:
+        Tensor: Image based relative point coordinates for sampling,
+            shape (N, P, 2)
+    """
+
+    assert (isinstance(img, tuple) and len(img) == 2) or \
+           (isinstance(img, torch.Tensor) and len(img.shape) == 4)
+
+    if isinstance(img, tuple):
+        h, w = img
+        scale = torch.tensor([w, h],
+                             dtype=torch.float,
+                             device=abs_img_points.device)
+        scale = scale.view(1, 1, 2)
+    else:
+        scale = get_shape_from_feature_map(img)
+
+    return abs_img_points / scale * spatial_scale
+
+
+def rel_roi_point_to_rel_img_point(rois,
+                                   rel_roi_points,
+                                   img,
+                                   spatial_scale=1.):
+    """Convert roi based relative point coordinates to image based absolute
+    point coordinates.
+
+    Args:
+        rois (Tensor): RoIs or BBoxes, shape (N, 4) or (N, 5)
+        rel_roi_points (Tensor): Point coordinates inside RoI, relative to
+            RoI, location, range (0, 1), shape (N, P, 2)
+        img (tuple/Tensor): (height, width) of image or feature map.
+        spatial_scale (float): Scale points by this factor. Default: 1.
+
+    Returns:
+        Tensor: Image based relative point coordinates for sampling,
+            shape (N, P, 2)
+    """
+
+    abs_img_point = rel_roi_point_to_abs_img_point(rois, rel_roi_points)
+    rel_img_point = abs_img_point_to_rel_img_point(abs_img_point, img,
+                                                   spatial_scale)
+
+    return rel_img_point
+
+
+def point_sample(input, points, align_corners=False, **kwargs):
+    """A wrapper around :func:`grid_sample` to support 3D point_coords tensors
+    Unlike :func:`torch.nn.functional.grid_sample` it assumes point_coords to
+    lie inside ``[0, 1] x [0, 1]`` square.
+
+    Args:
+        input (Tensor): Feature map, shape (N, C, H, W).
+        points (Tensor): Image based absolute point coordinates (normalized),
+            range [0, 1] x [0, 1], shape (N, P, 2) or (N, Hgrid, Wgrid, 2).
+        align_corners (bool): Whether align_corners. Default: False
+
+    Returns:
+        Tensor: Features of `point` on `input`, shape (N, C, P) or
+            (N, C, Hgrid, Wgrid).
+    """
+
+    add_dim = False
+    if points.dim() == 3:
+        add_dim = True
+        points = points.unsqueeze(2)
+    if is_in_onnx_export_without_custom_ops():
+        # If custom ops for onnx runtime not compiled use python
+        # implementation of grid_sample function to make onnx graph
+        # with supported nodes
+        output = bilinear_grid_sample(
+            input, denormalize(points), align_corners=align_corners)
+    else:
+        output = F.grid_sample(
+            input, denormalize(points), align_corners=align_corners, **kwargs)
+    if add_dim:
+        output = output.squeeze(3)
+    return output
+
+
+class SimpleRoIAlign(nn.Module):
+
+    def __init__(self, output_size, spatial_scale, aligned=True):
+        """Simple RoI align in PointRend, faster than standard RoIAlign.
+
+        Args:
+            output_size (tuple[int]): h, w
+            spatial_scale (float): scale the input boxes by this number
+            aligned (bool): if False, use the legacy implementation in
+                MMDetection, align_corners=True will be used in F.grid_sample.
+                If True, align the results more perfectly.
+        """
+
+        super(SimpleRoIAlign, self).__init__()
+        self.output_size = _pair(output_size)
+        self.spatial_scale = float(spatial_scale)
+        # to be consistent with other RoI ops
+        self.use_torchvision = False
+        self.aligned = aligned
+
+    def forward(self, features, rois):
+        num_imgs = features.size(0)
+        num_rois = rois.size(0)
+        rel_roi_points = generate_grid(
+            num_rois, self.output_size, device=rois.device)
+
+        if torch.onnx.is_in_onnx_export():
+            rel_img_points = rel_roi_point_to_rel_img_point(
+                rois, rel_roi_points, features, self.spatial_scale)
+            rel_img_points = rel_img_points.reshape(num_imgs, -1,
+                                                    *rel_img_points.shape[1:])
+            point_feats = point_sample(
+                features, rel_img_points, align_corners=not self.aligned)
+            point_feats = point_feats.transpose(1, 2)
+        else:
+            point_feats = []
+            for batch_ind in range(num_imgs):
+                # unravel batch dim
+                feat = features[batch_ind].unsqueeze(0)
+                inds = (rois[:, 0].long() == batch_ind)
+                if inds.any():
+                    rel_img_points = rel_roi_point_to_rel_img_point(
+                        rois[inds], rel_roi_points[inds], feat,
+                        self.spatial_scale).unsqueeze(0)
+                    point_feat = point_sample(
+                        feat, rel_img_points, align_corners=not self.aligned)
+                    point_feat = point_feat.squeeze(0).transpose(0, 1)
+                    point_feats.append(point_feat)
+
+            point_feats = torch.cat(point_feats, dim=0)
+
+        channels = features.size(1)
+        roi_feats = point_feats.reshape(num_rois, channels, *self.output_size)
+
+        return roi_feats
+
+    def __repr__(self):
+        format_str = self.__class__.__name__
+        format_str += '(output_size={}, spatial_scale={}'.format(
+            self.output_size, self.spatial_scale)
+        return format_str
diff --git a/annotator/uniformer/mmcv/ops/points_in_boxes.py b/annotator/uniformer/mmcv/ops/points_in_boxes.py
new file mode 100644
index 0000000000000000000000000000000000000000..4003173a53052161dbcd687a2fa1d755642fdab8
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/points_in_boxes.py
@@ -0,0 +1,133 @@
+import torch
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', [
+    'points_in_boxes_part_forward', 'points_in_boxes_cpu_forward',
+    'points_in_boxes_all_forward'
+])
+
+
+def points_in_boxes_part(points, boxes):
+    """Find the box in which each point is (CUDA).
+
+    Args:
+        points (torch.Tensor): [B, M, 3], [x, y, z] in LiDAR/DEPTH coordinate
+        boxes (torch.Tensor): [B, T, 7],
+            num_valid_boxes <= T, [x, y, z, x_size, y_size, z_size, rz] in
+            LiDAR/DEPTH coordinate, (x, y, z) is the bottom center
+
+    Returns:
+        box_idxs_of_pts (torch.Tensor): (B, M), default background = -1
+    """
+    assert points.shape[0] == boxes.shape[0], \
+        'Points and boxes should have the same batch size, ' \
+        f'but got {points.shape[0]} and {boxes.shape[0]}'
+    assert boxes.shape[2] == 7, \
+        'boxes dimension should be 7, ' \
+        f'but got unexpected shape {boxes.shape[2]}'
+    assert points.shape[2] == 3, \
+        'points dimension should be 3, ' \
+        f'but got unexpected shape {points.shape[2]}'
+    batch_size, num_points, _ = points.shape
+
+    box_idxs_of_pts = points.new_zeros((batch_size, num_points),
+                                       dtype=torch.int).fill_(-1)
+
+    # If manually put the tensor 'points' or 'boxes' on a device
+    # which is not the current device, some temporary variables
+    # will be created on the current device in the cuda op,
+    # and the output will be incorrect.
+    # Therefore, we force the current device to be the same
+    # as the device of the tensors if it was not.
+    # Please refer to https://github.com/open-mmlab/mmdetection3d/issues/305
+    # for the incorrect output before the fix.
+    points_device = points.get_device()
+    assert points_device == boxes.get_device(), \
+        'Points and boxes should be put on the same device'
+    if torch.cuda.current_device() != points_device:
+        torch.cuda.set_device(points_device)
+
+    ext_module.points_in_boxes_part_forward(boxes.contiguous(),
+                                            points.contiguous(),
+                                            box_idxs_of_pts)
+
+    return box_idxs_of_pts
+
+
+def points_in_boxes_cpu(points, boxes):
+    """Find all boxes in which each point is (CPU). The CPU version of
+    :meth:`points_in_boxes_all`.
+
+    Args:
+        points (torch.Tensor): [B, M, 3], [x, y, z] in
+            LiDAR/DEPTH coordinate
+        boxes (torch.Tensor): [B, T, 7],
+            num_valid_boxes <= T, [x, y, z, x_size, y_size, z_size, rz],
+            (x, y, z) is the bottom center.
+
+    Returns:
+        box_idxs_of_pts (torch.Tensor): (B, M, T), default background = 0.
+    """
+    assert points.shape[0] == boxes.shape[0], \
+        'Points and boxes should have the same batch size, ' \
+        f'but got {points.shape[0]} and {boxes.shape[0]}'
+    assert boxes.shape[2] == 7, \
+        'boxes dimension should be 7, ' \
+        f'but got unexpected shape {boxes.shape[2]}'
+    assert points.shape[2] == 3, \
+        'points dimension should be 3, ' \
+        f'but got unexpected shape {points.shape[2]}'
+    batch_size, num_points, _ = points.shape
+    num_boxes = boxes.shape[1]
+
+    point_indices = points.new_zeros((batch_size, num_boxes, num_points),
+                                     dtype=torch.int)
+    for b in range(batch_size):
+        ext_module.points_in_boxes_cpu_forward(boxes[b].float().contiguous(),
+                                               points[b].float().contiguous(),
+                                               point_indices[b])
+    point_indices = point_indices.transpose(1, 2)
+
+    return point_indices
+
+
+def points_in_boxes_all(points, boxes):
+    """Find all boxes in which each point is (CUDA).
+
+    Args:
+        points (torch.Tensor): [B, M, 3], [x, y, z] in LiDAR/DEPTH coordinate
+        boxes (torch.Tensor): [B, T, 7],
+            num_valid_boxes <= T, [x, y, z, x_size, y_size, z_size, rz],
+            (x, y, z) is the bottom center.
+
+    Returns:
+        box_idxs_of_pts (torch.Tensor): (B, M, T), default background = 0.
+    """
+    assert boxes.shape[0] == points.shape[0], \
+        'Points and boxes should have the same batch size, ' \
+        f'but got {boxes.shape[0]} and {boxes.shape[0]}'
+    assert boxes.shape[2] == 7, \
+        'boxes dimension should be 7, ' \
+        f'but got unexpected shape {boxes.shape[2]}'
+    assert points.shape[2] == 3, \
+        'points dimension should be 3, ' \
+        f'but got unexpected shape {points.shape[2]}'
+    batch_size, num_points, _ = points.shape
+    num_boxes = boxes.shape[1]
+
+    box_idxs_of_pts = points.new_zeros((batch_size, num_points, num_boxes),
+                                       dtype=torch.int).fill_(0)
+
+    # Same reason as line 25-32
+    points_device = points.get_device()
+    assert points_device == boxes.get_device(), \
+        'Points and boxes should be put on the same device'
+    if torch.cuda.current_device() != points_device:
+        torch.cuda.set_device(points_device)
+
+    ext_module.points_in_boxes_all_forward(boxes.contiguous(),
+                                           points.contiguous(),
+                                           box_idxs_of_pts)
+
+    return box_idxs_of_pts
diff --git a/annotator/uniformer/mmcv/ops/points_sampler.py b/annotator/uniformer/mmcv/ops/points_sampler.py
new file mode 100644
index 0000000000000000000000000000000000000000..a802a74fd6c3610d9ae178e6201f47423eca7ad1
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/points_sampler.py
@@ -0,0 +1,177 @@
+from typing import List
+
+import torch
+from torch import nn as nn
+
+from annotator.uniformer.mmcv.runner import force_fp32
+from .furthest_point_sample import (furthest_point_sample,
+                                    furthest_point_sample_with_dist)
+
+
+def calc_square_dist(point_feat_a, point_feat_b, norm=True):
+    """Calculating square distance between a and b.
+
+    Args:
+        point_feat_a (Tensor): (B, N, C) Feature vector of each point.
+        point_feat_b (Tensor): (B, M, C) Feature vector of each point.
+        norm (Bool, optional): Whether to normalize the distance.
+            Default: True.
+
+    Returns:
+        Tensor: (B, N, M) Distance between each pair points.
+    """
+    num_channel = point_feat_a.shape[-1]
+    # [bs, n, 1]
+    a_square = torch.sum(point_feat_a.unsqueeze(dim=2).pow(2), dim=-1)
+    # [bs, 1, m]
+    b_square = torch.sum(point_feat_b.unsqueeze(dim=1).pow(2), dim=-1)
+
+    corr_matrix = torch.matmul(point_feat_a, point_feat_b.transpose(1, 2))
+
+    dist = a_square + b_square - 2 * corr_matrix
+    if norm:
+        dist = torch.sqrt(dist) / num_channel
+    return dist
+
+
+def get_sampler_cls(sampler_type):
+    """Get the type and mode of points sampler.
+
+    Args:
+        sampler_type (str): The type of points sampler.
+            The valid value are "D-FPS", "F-FPS", or "FS".
+
+    Returns:
+        class: Points sampler type.
+    """
+    sampler_mappings = {
+        'D-FPS': DFPSSampler,
+        'F-FPS': FFPSSampler,
+        'FS': FSSampler,
+    }
+    try:
+        return sampler_mappings[sampler_type]
+    except KeyError:
+        raise KeyError(
+            f'Supported `sampler_type` are {sampler_mappings.keys()}, but got \
+                {sampler_type}')
+
+
+class PointsSampler(nn.Module):
+    """Points sampling.
+
+    Args:
+        num_point (list[int]): Number of sample points.
+        fps_mod_list (list[str], optional): Type of FPS method, valid mod
+            ['F-FPS', 'D-FPS', 'FS'], Default: ['D-FPS'].
+            F-FPS: using feature distances for FPS.
+            D-FPS: using Euclidean distances of points for FPS.
+            FS: using F-FPS and D-FPS simultaneously.
+        fps_sample_range_list (list[int], optional):
+            Range of points to apply FPS. Default: [-1].
+    """
+
+    def __init__(self,
+                 num_point: List[int],
+                 fps_mod_list: List[str] = ['D-FPS'],
+                 fps_sample_range_list: List[int] = [-1]):
+        super().__init__()
+        # FPS would be applied to different fps_mod in the list,
+        # so the length of the num_point should be equal to
+        # fps_mod_list and fps_sample_range_list.
+        assert len(num_point) == len(fps_mod_list) == len(
+            fps_sample_range_list)
+        self.num_point = num_point
+        self.fps_sample_range_list = fps_sample_range_list
+        self.samplers = nn.ModuleList()
+        for fps_mod in fps_mod_list:
+            self.samplers.append(get_sampler_cls(fps_mod)())
+        self.fp16_enabled = False
+
+    @force_fp32()
+    def forward(self, points_xyz, features):
+        """
+        Args:
+            points_xyz (Tensor): (B, N, 3) xyz coordinates of the features.
+            features (Tensor): (B, C, N) Descriptors of the features.
+
+        Returns:
+            Tensor: (B, npoint, sample_num) Indices of sampled points.
+        """
+        indices = []
+        last_fps_end_index = 0
+
+        for fps_sample_range, sampler, npoint in zip(
+                self.fps_sample_range_list, self.samplers, self.num_point):
+            assert fps_sample_range < points_xyz.shape[1]
+
+            if fps_sample_range == -1:
+                sample_points_xyz = points_xyz[:, last_fps_end_index:]
+                if features is not None:
+                    sample_features = features[:, :, last_fps_end_index:]
+                else:
+                    sample_features = None
+            else:
+                sample_points_xyz = \
+                    points_xyz[:, last_fps_end_index:fps_sample_range]
+                if features is not None:
+                    sample_features = features[:, :, last_fps_end_index:
+                                               fps_sample_range]
+                else:
+                    sample_features = None
+
+            fps_idx = sampler(sample_points_xyz.contiguous(), sample_features,
+                              npoint)
+
+            indices.append(fps_idx + last_fps_end_index)
+            last_fps_end_index += fps_sample_range
+        indices = torch.cat(indices, dim=1)
+
+        return indices
+
+
+class DFPSSampler(nn.Module):
+    """Using Euclidean distances of points for FPS."""
+
+    def __init__(self):
+        super().__init__()
+
+    def forward(self, points, features, npoint):
+        """Sampling points with D-FPS."""
+        fps_idx = furthest_point_sample(points.contiguous(), npoint)
+        return fps_idx
+
+
+class FFPSSampler(nn.Module):
+    """Using feature distances for FPS."""
+
+    def __init__(self):
+        super().__init__()
+
+    def forward(self, points, features, npoint):
+        """Sampling points with F-FPS."""
+        assert features is not None, \
+            'feature input to FFPS_Sampler should not be None'
+        features_for_fps = torch.cat([points, features.transpose(1, 2)], dim=2)
+        features_dist = calc_square_dist(
+            features_for_fps, features_for_fps, norm=False)
+        fps_idx = furthest_point_sample_with_dist(features_dist, npoint)
+        return fps_idx
+
+
+class FSSampler(nn.Module):
+    """Using F-FPS and D-FPS simultaneously."""
+
+    def __init__(self):
+        super().__init__()
+
+    def forward(self, points, features, npoint):
+        """Sampling points with FS_Sampling."""
+        assert features is not None, \
+            'feature input to FS_Sampler should not be None'
+        ffps_sampler = FFPSSampler()
+        dfps_sampler = DFPSSampler()
+        fps_idx_ffps = ffps_sampler(points, features, npoint)
+        fps_idx_dfps = dfps_sampler(points, features, npoint)
+        fps_idx = torch.cat([fps_idx_ffps, fps_idx_dfps], dim=1)
+        return fps_idx
diff --git a/annotator/uniformer/mmcv/ops/psa_mask.py b/annotator/uniformer/mmcv/ops/psa_mask.py
new file mode 100644
index 0000000000000000000000000000000000000000..cdf14e62b50e8d4dd6856c94333c703bcc4c9ab6
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/psa_mask.py
@@ -0,0 +1,92 @@
+# Modified from https://github.com/hszhao/semseg/blob/master/lib/psa
+from torch import nn
+from torch.autograd import Function
+from torch.nn.modules.utils import _pair
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext',
+                                 ['psamask_forward', 'psamask_backward'])
+
+
+class PSAMaskFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, psa_type, mask_size):
+        return g.op(
+            'mmcv::MMCVPSAMask',
+            input,
+            psa_type_i=psa_type,
+            mask_size_i=mask_size)
+
+    @staticmethod
+    def forward(ctx, input, psa_type, mask_size):
+        ctx.psa_type = psa_type
+        ctx.mask_size = _pair(mask_size)
+        ctx.save_for_backward(input)
+
+        h_mask, w_mask = ctx.mask_size
+        batch_size, channels, h_feature, w_feature = input.size()
+        assert channels == h_mask * w_mask
+        output = input.new_zeros(
+            (batch_size, h_feature * w_feature, h_feature, w_feature))
+
+        ext_module.psamask_forward(
+            input,
+            output,
+            psa_type=psa_type,
+            num_=batch_size,
+            h_feature=h_feature,
+            w_feature=w_feature,
+            h_mask=h_mask,
+            w_mask=w_mask,
+            half_h_mask=(h_mask - 1) // 2,
+            half_w_mask=(w_mask - 1) // 2)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        input = ctx.saved_tensors[0]
+        psa_type = ctx.psa_type
+        h_mask, w_mask = ctx.mask_size
+        batch_size, channels, h_feature, w_feature = input.size()
+        grad_input = grad_output.new_zeros(
+            (batch_size, channels, h_feature, w_feature))
+        ext_module.psamask_backward(
+            grad_output,
+            grad_input,
+            psa_type=psa_type,
+            num_=batch_size,
+            h_feature=h_feature,
+            w_feature=w_feature,
+            h_mask=h_mask,
+            w_mask=w_mask,
+            half_h_mask=(h_mask - 1) // 2,
+            half_w_mask=(w_mask - 1) // 2)
+        return grad_input, None, None, None
+
+
+psa_mask = PSAMaskFunction.apply
+
+
+class PSAMask(nn.Module):
+
+    def __init__(self, psa_type, mask_size=None):
+        super(PSAMask, self).__init__()
+        assert psa_type in ['collect', 'distribute']
+        if psa_type == 'collect':
+            psa_type_enum = 0
+        else:
+            psa_type_enum = 1
+        self.psa_type_enum = psa_type_enum
+        self.mask_size = mask_size
+        self.psa_type = psa_type
+
+    def forward(self, input):
+        return psa_mask(input, self.psa_type_enum, self.mask_size)
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(psa_type={self.psa_type}, '
+        s += f'mask_size={self.mask_size})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/roi_align.py b/annotator/uniformer/mmcv/ops/roi_align.py
new file mode 100644
index 0000000000000000000000000000000000000000..0755aefc66e67233ceae0f4b77948301c443e9fb
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/roi_align.py
@@ -0,0 +1,223 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.utils import _pair
+
+from ..utils import deprecated_api_warning, ext_loader
+
+ext_module = ext_loader.load_ext('_ext',
+                                 ['roi_align_forward', 'roi_align_backward'])
+
+
+class RoIAlignFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, rois, output_size, spatial_scale, sampling_ratio,
+                 pool_mode, aligned):
+        from ..onnx import is_custom_op_loaded
+        has_custom_op = is_custom_op_loaded()
+        if has_custom_op:
+            return g.op(
+                'mmcv::MMCVRoiAlign',
+                input,
+                rois,
+                output_height_i=output_size[0],
+                output_width_i=output_size[1],
+                spatial_scale_f=spatial_scale,
+                sampling_ratio_i=sampling_ratio,
+                mode_s=pool_mode,
+                aligned_i=aligned)
+        else:
+            from torch.onnx.symbolic_opset9 import sub, squeeze
+            from torch.onnx.symbolic_helper import _slice_helper
+            from torch.onnx import TensorProtoDataType
+            # batch_indices = rois[:, 0].long()
+            batch_indices = _slice_helper(
+                g, rois, axes=[1], starts=[0], ends=[1])
+            batch_indices = squeeze(g, batch_indices, 1)
+            batch_indices = g.op(
+                'Cast', batch_indices, to_i=TensorProtoDataType.INT64)
+            # rois = rois[:, 1:]
+            rois = _slice_helper(g, rois, axes=[1], starts=[1], ends=[5])
+            if aligned:
+                # rois -= 0.5/spatial_scale
+                aligned_offset = g.op(
+                    'Constant',
+                    value_t=torch.tensor([0.5 / spatial_scale],
+                                         dtype=torch.float32))
+                rois = sub(g, rois, aligned_offset)
+            # roi align
+            return g.op(
+                'RoiAlign',
+                input,
+                rois,
+                batch_indices,
+                output_height_i=output_size[0],
+                output_width_i=output_size[1],
+                spatial_scale_f=spatial_scale,
+                sampling_ratio_i=max(0, sampling_ratio),
+                mode_s=pool_mode)
+
+    @staticmethod
+    def forward(ctx,
+                input,
+                rois,
+                output_size,
+                spatial_scale=1.0,
+                sampling_ratio=0,
+                pool_mode='avg',
+                aligned=True):
+        ctx.output_size = _pair(output_size)
+        ctx.spatial_scale = spatial_scale
+        ctx.sampling_ratio = sampling_ratio
+        assert pool_mode in ('max', 'avg')
+        ctx.pool_mode = 0 if pool_mode == 'max' else 1
+        ctx.aligned = aligned
+        ctx.input_shape = input.size()
+
+        assert rois.size(1) == 5, 'RoI must be (idx, x1, y1, x2, y2)!'
+
+        output_shape = (rois.size(0), input.size(1), ctx.output_size[0],
+                        ctx.output_size[1])
+        output = input.new_zeros(output_shape)
+        if ctx.pool_mode == 0:
+            argmax_y = input.new_zeros(output_shape)
+            argmax_x = input.new_zeros(output_shape)
+        else:
+            argmax_y = input.new_zeros(0)
+            argmax_x = input.new_zeros(0)
+
+        ext_module.roi_align_forward(
+            input,
+            rois,
+            output,
+            argmax_y,
+            argmax_x,
+            aligned_height=ctx.output_size[0],
+            aligned_width=ctx.output_size[1],
+            spatial_scale=ctx.spatial_scale,
+            sampling_ratio=ctx.sampling_ratio,
+            pool_mode=ctx.pool_mode,
+            aligned=ctx.aligned)
+
+        ctx.save_for_backward(rois, argmax_y, argmax_x)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        rois, argmax_y, argmax_x = ctx.saved_tensors
+        grad_input = grad_output.new_zeros(ctx.input_shape)
+        # complex head architecture may cause grad_output uncontiguous.
+        grad_output = grad_output.contiguous()
+        ext_module.roi_align_backward(
+            grad_output,
+            rois,
+            argmax_y,
+            argmax_x,
+            grad_input,
+            aligned_height=ctx.output_size[0],
+            aligned_width=ctx.output_size[1],
+            spatial_scale=ctx.spatial_scale,
+            sampling_ratio=ctx.sampling_ratio,
+            pool_mode=ctx.pool_mode,
+            aligned=ctx.aligned)
+        return grad_input, None, None, None, None, None, None
+
+
+roi_align = RoIAlignFunction.apply
+
+
+class RoIAlign(nn.Module):
+    """RoI align pooling layer.
+
+    Args:
+        output_size (tuple): h, w
+        spatial_scale (float): scale the input boxes by this number
+        sampling_ratio (int): number of inputs samples to take for each
+            output sample. 0 to take samples densely for current models.
+        pool_mode (str, 'avg' or 'max'): pooling mode in each bin.
+        aligned (bool): if False, use the legacy implementation in
+            MMDetection. If True, align the results more perfectly.
+        use_torchvision (bool): whether to use roi_align from torchvision.
+
+    Note:
+        The implementation of RoIAlign when aligned=True is modified from
+        https://github.com/facebookresearch/detectron2/
+
+        The meaning of aligned=True:
+
+        Given a continuous coordinate c, its two neighboring pixel
+        indices (in our pixel model) are computed by floor(c - 0.5) and
+        ceil(c - 0.5). For example, c=1.3 has pixel neighbors with discrete
+        indices [0] and [1] (which are sampled from the underlying signal
+        at continuous coordinates 0.5 and 1.5). But the original roi_align
+        (aligned=False) does not subtract the 0.5 when computing
+        neighboring pixel indices and therefore it uses pixels with a
+        slightly incorrect alignment (relative to our pixel model) when
+        performing bilinear interpolation.
+
+        With `aligned=True`,
+        we first appropriately scale the ROI and then shift it by -0.5
+        prior to calling roi_align. This produces the correct neighbors;
+
+        The difference does not make a difference to the model's
+        performance if ROIAlign is used together with conv layers.
+    """
+
+    @deprecated_api_warning(
+        {
+            'out_size': 'output_size',
+            'sample_num': 'sampling_ratio'
+        },
+        cls_name='RoIAlign')
+    def __init__(self,
+                 output_size,
+                 spatial_scale=1.0,
+                 sampling_ratio=0,
+                 pool_mode='avg',
+                 aligned=True,
+                 use_torchvision=False):
+        super(RoIAlign, self).__init__()
+
+        self.output_size = _pair(output_size)
+        self.spatial_scale = float(spatial_scale)
+        self.sampling_ratio = int(sampling_ratio)
+        self.pool_mode = pool_mode
+        self.aligned = aligned
+        self.use_torchvision = use_torchvision
+
+    def forward(self, input, rois):
+        """
+        Args:
+            input: NCHW images
+            rois: Bx5 boxes. First column is the index into N.\
+                The other 4 columns are xyxy.
+        """
+        if self.use_torchvision:
+            from torchvision.ops import roi_align as tv_roi_align
+            if 'aligned' in tv_roi_align.__code__.co_varnames:
+                return tv_roi_align(input, rois, self.output_size,
+                                    self.spatial_scale, self.sampling_ratio,
+                                    self.aligned)
+            else:
+                if self.aligned:
+                    rois -= rois.new_tensor([0.] +
+                                            [0.5 / self.spatial_scale] * 4)
+                return tv_roi_align(input, rois, self.output_size,
+                                    self.spatial_scale, self.sampling_ratio)
+        else:
+            return roi_align(input, rois, self.output_size, self.spatial_scale,
+                             self.sampling_ratio, self.pool_mode, self.aligned)
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(output_size={self.output_size}, '
+        s += f'spatial_scale={self.spatial_scale}, '
+        s += f'sampling_ratio={self.sampling_ratio}, '
+        s += f'pool_mode={self.pool_mode}, '
+        s += f'aligned={self.aligned}, '
+        s += f'use_torchvision={self.use_torchvision})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/roi_align_rotated.py b/annotator/uniformer/mmcv/ops/roi_align_rotated.py
new file mode 100644
index 0000000000000000000000000000000000000000..0ce4961a3555d4da8bc3e32f1f7d5ad50036587d
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/roi_align_rotated.py
@@ -0,0 +1,177 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch.nn as nn
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['roi_align_rotated_forward', 'roi_align_rotated_backward'])
+
+
+class RoIAlignRotatedFunction(Function):
+
+    @staticmethod
+    def symbolic(g, features, rois, out_size, spatial_scale, sample_num,
+                 aligned, clockwise):
+        if isinstance(out_size, int):
+            out_h = out_size
+            out_w = out_size
+        elif isinstance(out_size, tuple):
+            assert len(out_size) == 2
+            assert isinstance(out_size[0], int)
+            assert isinstance(out_size[1], int)
+            out_h, out_w = out_size
+        else:
+            raise TypeError(
+                '"out_size" must be an integer or tuple of integers')
+        return g.op(
+            'mmcv::MMCVRoIAlignRotated',
+            features,
+            rois,
+            output_height_i=out_h,
+            output_width_i=out_h,
+            spatial_scale_f=spatial_scale,
+            sampling_ratio_i=sample_num,
+            aligned_i=aligned,
+            clockwise_i=clockwise)
+
+    @staticmethod
+    def forward(ctx,
+                features,
+                rois,
+                out_size,
+                spatial_scale,
+                sample_num=0,
+                aligned=True,
+                clockwise=False):
+        if isinstance(out_size, int):
+            out_h = out_size
+            out_w = out_size
+        elif isinstance(out_size, tuple):
+            assert len(out_size) == 2
+            assert isinstance(out_size[0], int)
+            assert isinstance(out_size[1], int)
+            out_h, out_w = out_size
+        else:
+            raise TypeError(
+                '"out_size" must be an integer or tuple of integers')
+        ctx.spatial_scale = spatial_scale
+        ctx.sample_num = sample_num
+        ctx.aligned = aligned
+        ctx.clockwise = clockwise
+        ctx.save_for_backward(rois)
+        ctx.feature_size = features.size()
+
+        batch_size, num_channels, data_height, data_width = features.size()
+        num_rois = rois.size(0)
+
+        output = features.new_zeros(num_rois, num_channels, out_h, out_w)
+        ext_module.roi_align_rotated_forward(
+            features,
+            rois,
+            output,
+            pooled_height=out_h,
+            pooled_width=out_w,
+            spatial_scale=spatial_scale,
+            sample_num=sample_num,
+            aligned=aligned,
+            clockwise=clockwise)
+        return output
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        feature_size = ctx.feature_size
+        spatial_scale = ctx.spatial_scale
+        aligned = ctx.aligned
+        clockwise = ctx.clockwise
+        sample_num = ctx.sample_num
+        rois = ctx.saved_tensors[0]
+        assert feature_size is not None
+        batch_size, num_channels, data_height, data_width = feature_size
+
+        out_w = grad_output.size(3)
+        out_h = grad_output.size(2)
+
+        grad_input = grad_rois = None
+
+        if ctx.needs_input_grad[0]:
+            grad_input = rois.new_zeros(batch_size, num_channels, data_height,
+                                        data_width)
+            ext_module.roi_align_rotated_backward(
+                grad_output.contiguous(),
+                rois,
+                grad_input,
+                pooled_height=out_h,
+                pooled_width=out_w,
+                spatial_scale=spatial_scale,
+                sample_num=sample_num,
+                aligned=aligned,
+                clockwise=clockwise)
+        return grad_input, grad_rois, None, None, None, None, None
+
+
+roi_align_rotated = RoIAlignRotatedFunction.apply
+
+
+class RoIAlignRotated(nn.Module):
+    """RoI align pooling layer for rotated proposals.
+
+    It accepts a feature map of shape (N, C, H, W) and rois with shape
+    (n, 6) with each roi decoded as (batch_index, center_x, center_y,
+    w, h, angle). The angle is in radian.
+
+    Args:
+        out_size (tuple): h, w
+        spatial_scale (float): scale the input boxes by this number
+        sample_num (int): number of inputs samples to take for each
+            output sample. 0 to take samples densely for current models.
+        aligned (bool): if False, use the legacy implementation in
+            MMDetection. If True, align the results more perfectly.
+            Default: True.
+        clockwise (bool): If True, the angle in each proposal follows a
+            clockwise fashion in image space, otherwise, the angle is
+            counterclockwise. Default: False.
+
+    Note:
+        The implementation of RoIAlign when aligned=True is modified from
+        https://github.com/facebookresearch/detectron2/
+
+        The meaning of aligned=True:
+
+        Given a continuous coordinate c, its two neighboring pixel
+        indices (in our pixel model) are computed by floor(c - 0.5) and
+        ceil(c - 0.5). For example, c=1.3 has pixel neighbors with discrete
+        indices [0] and [1] (which are sampled from the underlying signal
+        at continuous coordinates 0.5 and 1.5). But the original roi_align
+        (aligned=False) does not subtract the 0.5 when computing
+        neighboring pixel indices and therefore it uses pixels with a
+        slightly incorrect alignment (relative to our pixel model) when
+        performing bilinear interpolation.
+
+        With `aligned=True`,
+        we first appropriately scale the ROI and then shift it by -0.5
+        prior to calling roi_align. This produces the correct neighbors;
+
+        The difference does not make a difference to the model's
+        performance if ROIAlign is used together with conv layers.
+    """
+
+    def __init__(self,
+                 out_size,
+                 spatial_scale,
+                 sample_num=0,
+                 aligned=True,
+                 clockwise=False):
+        super(RoIAlignRotated, self).__init__()
+
+        self.out_size = out_size
+        self.spatial_scale = float(spatial_scale)
+        self.sample_num = int(sample_num)
+        self.aligned = aligned
+        self.clockwise = clockwise
+
+    def forward(self, features, rois):
+        return RoIAlignRotatedFunction.apply(features, rois, self.out_size,
+                                             self.spatial_scale,
+                                             self.sample_num, self.aligned,
+                                             self.clockwise)
diff --git a/annotator/uniformer/mmcv/ops/roi_pool.py b/annotator/uniformer/mmcv/ops/roi_pool.py
new file mode 100644
index 0000000000000000000000000000000000000000..d339d8f2941eabc1cbe181a9c6c5ab5ff4ff4e5f
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/roi_pool.py
@@ -0,0 +1,86 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.utils import _pair
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext',
+                                 ['roi_pool_forward', 'roi_pool_backward'])
+
+
+class RoIPoolFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, rois, output_size, spatial_scale):
+        return g.op(
+            'MaxRoiPool',
+            input,
+            rois,
+            pooled_shape_i=output_size,
+            spatial_scale_f=spatial_scale)
+
+    @staticmethod
+    def forward(ctx, input, rois, output_size, spatial_scale=1.0):
+        ctx.output_size = _pair(output_size)
+        ctx.spatial_scale = spatial_scale
+        ctx.input_shape = input.size()
+
+        assert rois.size(1) == 5, 'RoI must be (idx, x1, y1, x2, y2)!'
+
+        output_shape = (rois.size(0), input.size(1), ctx.output_size[0],
+                        ctx.output_size[1])
+        output = input.new_zeros(output_shape)
+        argmax = input.new_zeros(output_shape, dtype=torch.int)
+
+        ext_module.roi_pool_forward(
+            input,
+            rois,
+            output,
+            argmax,
+            pooled_height=ctx.output_size[0],
+            pooled_width=ctx.output_size[1],
+            spatial_scale=ctx.spatial_scale)
+
+        ctx.save_for_backward(rois, argmax)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(ctx, grad_output):
+        rois, argmax = ctx.saved_tensors
+        grad_input = grad_output.new_zeros(ctx.input_shape)
+
+        ext_module.roi_pool_backward(
+            grad_output,
+            rois,
+            argmax,
+            grad_input,
+            pooled_height=ctx.output_size[0],
+            pooled_width=ctx.output_size[1],
+            spatial_scale=ctx.spatial_scale)
+
+        return grad_input, None, None, None
+
+
+roi_pool = RoIPoolFunction.apply
+
+
+class RoIPool(nn.Module):
+
+    def __init__(self, output_size, spatial_scale=1.0):
+        super(RoIPool, self).__init__()
+
+        self.output_size = _pair(output_size)
+        self.spatial_scale = float(spatial_scale)
+
+    def forward(self, input, rois):
+        return roi_pool(input, rois, self.output_size, self.spatial_scale)
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'(output_size={self.output_size}, '
+        s += f'spatial_scale={self.spatial_scale})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/roiaware_pool3d.py b/annotator/uniformer/mmcv/ops/roiaware_pool3d.py
new file mode 100644
index 0000000000000000000000000000000000000000..291b0e5a9b692492c7d7e495ea639c46042e2f18
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/roiaware_pool3d.py
@@ -0,0 +1,114 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch import nn as nn
+from torch.autograd import Function
+
+import annotator.uniformer.mmcv as mmcv
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['roiaware_pool3d_forward', 'roiaware_pool3d_backward'])
+
+
+class RoIAwarePool3d(nn.Module):
+    """Encode the geometry-specific features of each 3D proposal.
+
+    Please refer to `PartA2 <https://arxiv.org/pdf/1907.03670.pdf>`_ for more
+    details.
+
+    Args:
+        out_size (int or tuple): The size of output features. n or
+            [n1, n2, n3].
+        max_pts_per_voxel (int, optional): The maximum number of points per
+            voxel. Default: 128.
+        mode (str, optional): Pooling method of RoIAware, 'max' or 'avg'.
+            Default: 'max'.
+    """
+
+    def __init__(self, out_size, max_pts_per_voxel=128, mode='max'):
+        super().__init__()
+
+        self.out_size = out_size
+        self.max_pts_per_voxel = max_pts_per_voxel
+        assert mode in ['max', 'avg']
+        pool_mapping = {'max': 0, 'avg': 1}
+        self.mode = pool_mapping[mode]
+
+    def forward(self, rois, pts, pts_feature):
+        """
+        Args:
+            rois (torch.Tensor): [N, 7], in LiDAR coordinate,
+                (x, y, z) is the bottom center of rois.
+            pts (torch.Tensor): [npoints, 3], coordinates of input points.
+            pts_feature (torch.Tensor): [npoints, C], features of input points.
+
+        Returns:
+            pooled_features (torch.Tensor): [N, out_x, out_y, out_z, C]
+        """
+
+        return RoIAwarePool3dFunction.apply(rois, pts, pts_feature,
+                                            self.out_size,
+                                            self.max_pts_per_voxel, self.mode)
+
+
+class RoIAwarePool3dFunction(Function):
+
+    @staticmethod
+    def forward(ctx, rois, pts, pts_feature, out_size, max_pts_per_voxel,
+                mode):
+        """
+        Args:
+            rois (torch.Tensor): [N, 7], in LiDAR coordinate,
+                (x, y, z) is the bottom center of rois.
+            pts (torch.Tensor): [npoints, 3], coordinates of input points.
+            pts_feature (torch.Tensor): [npoints, C], features of input points.
+            out_size (int or tuple): The size of output features. n or
+                [n1, n2, n3].
+            max_pts_per_voxel (int): The maximum number of points per voxel.
+                Default: 128.
+            mode (int): Pooling method of RoIAware, 0 (max pool) or 1 (average
+                pool).
+
+        Returns:
+            pooled_features (torch.Tensor): [N, out_x, out_y, out_z, C], output
+                pooled features.
+        """
+
+        if isinstance(out_size, int):
+            out_x = out_y = out_z = out_size
+        else:
+            assert len(out_size) == 3
+            assert mmcv.is_tuple_of(out_size, int)
+            out_x, out_y, out_z = out_size
+
+        num_rois = rois.shape[0]
+        num_channels = pts_feature.shape[-1]
+        num_pts = pts.shape[0]
+
+        pooled_features = pts_feature.new_zeros(
+            (num_rois, out_x, out_y, out_z, num_channels))
+        argmax = pts_feature.new_zeros(
+            (num_rois, out_x, out_y, out_z, num_channels), dtype=torch.int)
+        pts_idx_of_voxels = pts_feature.new_zeros(
+            (num_rois, out_x, out_y, out_z, max_pts_per_voxel),
+            dtype=torch.int)
+
+        ext_module.roiaware_pool3d_forward(rois, pts, pts_feature, argmax,
+                                           pts_idx_of_voxels, pooled_features,
+                                           mode)
+
+        ctx.roiaware_pool3d_for_backward = (pts_idx_of_voxels, argmax, mode,
+                                            num_pts, num_channels)
+        return pooled_features
+
+    @staticmethod
+    def backward(ctx, grad_out):
+        ret = ctx.roiaware_pool3d_for_backward
+        pts_idx_of_voxels, argmax, mode, num_pts, num_channels = ret
+
+        grad_in = grad_out.new_zeros((num_pts, num_channels))
+        ext_module.roiaware_pool3d_backward(pts_idx_of_voxels, argmax,
+                                            grad_out.contiguous(), grad_in,
+                                            mode)
+
+        return None, None, grad_in, None, None, None
diff --git a/annotator/uniformer/mmcv/ops/roipoint_pool3d.py b/annotator/uniformer/mmcv/ops/roipoint_pool3d.py
new file mode 100644
index 0000000000000000000000000000000000000000..0a21412c0728431c04b84245bc2e3109eea9aefc
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/roipoint_pool3d.py
@@ -0,0 +1,77 @@
+from torch import nn as nn
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['roipoint_pool3d_forward'])
+
+
+class RoIPointPool3d(nn.Module):
+    """Encode the geometry-specific features of each 3D proposal.
+
+    Please refer to `Paper of PartA2 <https://arxiv.org/pdf/1907.03670.pdf>`_
+    for more details.
+
+    Args:
+        num_sampled_points (int, optional): Number of samples in each roi.
+            Default: 512.
+    """
+
+    def __init__(self, num_sampled_points=512):
+        super().__init__()
+        self.num_sampled_points = num_sampled_points
+
+    def forward(self, points, point_features, boxes3d):
+        """
+        Args:
+            points (torch.Tensor): Input points whose shape is (B, N, C).
+            point_features (torch.Tensor): Features of input points whose shape
+                is (B, N, C).
+            boxes3d (B, M, 7), Input bounding boxes whose shape is (B, M, 7).
+
+        Returns:
+            pooled_features (torch.Tensor): The output pooled features whose
+                shape is (B, M, 512, 3 + C).
+            pooled_empty_flag (torch.Tensor): Empty flag whose shape is (B, M).
+        """
+        return RoIPointPool3dFunction.apply(points, point_features, boxes3d,
+                                            self.num_sampled_points)
+
+
+class RoIPointPool3dFunction(Function):
+
+    @staticmethod
+    def forward(ctx, points, point_features, boxes3d, num_sampled_points=512):
+        """
+        Args:
+            points (torch.Tensor): Input points whose shape is (B, N, C).
+            point_features (torch.Tensor): Features of input points whose shape
+                is (B, N, C).
+            boxes3d (B, M, 7), Input bounding boxes whose shape is (B, M, 7).
+            num_sampled_points (int, optional): The num of sampled points.
+                Default: 512.
+
+        Returns:
+            pooled_features (torch.Tensor): The output pooled features whose
+                shape is (B, M, 512, 3 + C).
+            pooled_empty_flag (torch.Tensor): Empty flag whose shape is (B, M).
+        """
+        assert len(points.shape) == 3 and points.shape[2] == 3
+        batch_size, boxes_num, feature_len = points.shape[0], boxes3d.shape[
+            1], point_features.shape[2]
+        pooled_boxes3d = boxes3d.view(batch_size, -1, 7)
+        pooled_features = point_features.new_zeros(
+            (batch_size, boxes_num, num_sampled_points, 3 + feature_len))
+        pooled_empty_flag = point_features.new_zeros(
+            (batch_size, boxes_num)).int()
+
+        ext_module.roipoint_pool3d_forward(points.contiguous(),
+                                           pooled_boxes3d.contiguous(),
+                                           point_features.contiguous(),
+                                           pooled_features, pooled_empty_flag)
+
+        return pooled_features, pooled_empty_flag
+
+    @staticmethod
+    def backward(ctx, grad_out):
+        raise NotImplementedError
diff --git a/annotator/uniformer/mmcv/ops/saconv.py b/annotator/uniformer/mmcv/ops/saconv.py
new file mode 100644
index 0000000000000000000000000000000000000000..b4ee3978e097fca422805db4e31ae481006d7971
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/saconv.py
@@ -0,0 +1,145 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from annotator.uniformer.mmcv.cnn import CONV_LAYERS, ConvAWS2d, constant_init
+from annotator.uniformer.mmcv.ops.deform_conv import deform_conv2d
+from annotator.uniformer.mmcv.utils import TORCH_VERSION, digit_version
+
+
+@CONV_LAYERS.register_module(name='SAC')
+class SAConv2d(ConvAWS2d):
+    """SAC (Switchable Atrous Convolution)
+
+    This is an implementation of SAC in DetectoRS
+    (https://arxiv.org/pdf/2006.02334.pdf).
+
+    Args:
+        in_channels (int): Number of channels in the input image
+        out_channels (int): Number of channels produced by the convolution
+        kernel_size (int or tuple): Size of the convolving kernel
+        stride (int or tuple, optional): Stride of the convolution. Default: 1
+        padding (int or tuple, optional): Zero-padding added to both sides of
+            the input. Default: 0
+        padding_mode (string, optional): ``'zeros'``, ``'reflect'``,
+            ``'replicate'`` or ``'circular'``. Default: ``'zeros'``
+        dilation (int or tuple, optional): Spacing between kernel elements.
+            Default: 1
+        groups (int, optional): Number of blocked connections from input
+            channels to output channels. Default: 1
+        bias (bool, optional): If ``True``, adds a learnable bias to the
+            output. Default: ``True``
+        use_deform: If ``True``, replace convolution with deformable
+            convolution. Default: ``False``.
+    """
+
+    def __init__(self,
+                 in_channels,
+                 out_channels,
+                 kernel_size,
+                 stride=1,
+                 padding=0,
+                 dilation=1,
+                 groups=1,
+                 bias=True,
+                 use_deform=False):
+        super().__init__(
+            in_channels,
+            out_channels,
+            kernel_size,
+            stride=stride,
+            padding=padding,
+            dilation=dilation,
+            groups=groups,
+            bias=bias)
+        self.use_deform = use_deform
+        self.switch = nn.Conv2d(
+            self.in_channels, 1, kernel_size=1, stride=stride, bias=True)
+        self.weight_diff = nn.Parameter(torch.Tensor(self.weight.size()))
+        self.pre_context = nn.Conv2d(
+            self.in_channels, self.in_channels, kernel_size=1, bias=True)
+        self.post_context = nn.Conv2d(
+            self.out_channels, self.out_channels, kernel_size=1, bias=True)
+        if self.use_deform:
+            self.offset_s = nn.Conv2d(
+                self.in_channels,
+                18,
+                kernel_size=3,
+                padding=1,
+                stride=stride,
+                bias=True)
+            self.offset_l = nn.Conv2d(
+                self.in_channels,
+                18,
+                kernel_size=3,
+                padding=1,
+                stride=stride,
+                bias=True)
+        self.init_weights()
+
+    def init_weights(self):
+        constant_init(self.switch, 0, bias=1)
+        self.weight_diff.data.zero_()
+        constant_init(self.pre_context, 0)
+        constant_init(self.post_context, 0)
+        if self.use_deform:
+            constant_init(self.offset_s, 0)
+            constant_init(self.offset_l, 0)
+
+    def forward(self, x):
+        # pre-context
+        avg_x = F.adaptive_avg_pool2d(x, output_size=1)
+        avg_x = self.pre_context(avg_x)
+        avg_x = avg_x.expand_as(x)
+        x = x + avg_x
+        # switch
+        avg_x = F.pad(x, pad=(2, 2, 2, 2), mode='reflect')
+        avg_x = F.avg_pool2d(avg_x, kernel_size=5, stride=1, padding=0)
+        switch = self.switch(avg_x)
+        # sac
+        weight = self._get_weight(self.weight)
+        zero_bias = torch.zeros(
+            self.out_channels, device=weight.device, dtype=weight.dtype)
+
+        if self.use_deform:
+            offset = self.offset_s(avg_x)
+            out_s = deform_conv2d(x, offset, weight, self.stride, self.padding,
+                                  self.dilation, self.groups, 1)
+        else:
+            if (TORCH_VERSION == 'parrots'
+                    or digit_version(TORCH_VERSION) < digit_version('1.5.0')):
+                out_s = super().conv2d_forward(x, weight)
+            elif digit_version(TORCH_VERSION) >= digit_version('1.8.0'):
+                # bias is a required argument of _conv_forward in torch 1.8.0
+                out_s = super()._conv_forward(x, weight, zero_bias)
+            else:
+                out_s = super()._conv_forward(x, weight)
+        ori_p = self.padding
+        ori_d = self.dilation
+        self.padding = tuple(3 * p for p in self.padding)
+        self.dilation = tuple(3 * d for d in self.dilation)
+        weight = weight + self.weight_diff
+        if self.use_deform:
+            offset = self.offset_l(avg_x)
+            out_l = deform_conv2d(x, offset, weight, self.stride, self.padding,
+                                  self.dilation, self.groups, 1)
+        else:
+            if (TORCH_VERSION == 'parrots'
+                    or digit_version(TORCH_VERSION) < digit_version('1.5.0')):
+                out_l = super().conv2d_forward(x, weight)
+            elif digit_version(TORCH_VERSION) >= digit_version('1.8.0'):
+                # bias is a required argument of _conv_forward in torch 1.8.0
+                out_l = super()._conv_forward(x, weight, zero_bias)
+            else:
+                out_l = super()._conv_forward(x, weight)
+
+        out = switch * out_s + (1 - switch) * out_l
+        self.padding = ori_p
+        self.dilation = ori_d
+        # post-context
+        avg_x = F.adaptive_avg_pool2d(out, output_size=1)
+        avg_x = self.post_context(avg_x)
+        avg_x = avg_x.expand_as(out)
+        out = out + avg_x
+        return out
diff --git a/annotator/uniformer/mmcv/ops/scatter_points.py b/annotator/uniformer/mmcv/ops/scatter_points.py
new file mode 100644
index 0000000000000000000000000000000000000000..2b8aa4169e9f6ca4a6f845ce17d6d1e4db416bb8
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/scatter_points.py
@@ -0,0 +1,135 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch import nn
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext',
+    ['dynamic_point_to_voxel_forward', 'dynamic_point_to_voxel_backward'])
+
+
+class _DynamicScatter(Function):
+
+    @staticmethod
+    def forward(ctx, feats, coors, reduce_type='max'):
+        """convert kitti points(N, >=3) to voxels.
+
+        Args:
+            feats (torch.Tensor): [N, C]. Points features to be reduced
+                into voxels.
+            coors (torch.Tensor): [N, ndim]. Corresponding voxel coordinates
+                (specifically multi-dim voxel index) of each points.
+            reduce_type (str, optional): Reduce op. support 'max', 'sum' and
+                'mean'. Default: 'max'.
+
+        Returns:
+            voxel_feats (torch.Tensor): [M, C]. Reduced features, input
+                features that shares the same voxel coordinates are reduced to
+                one row.
+            voxel_coors (torch.Tensor): [M, ndim]. Voxel coordinates.
+        """
+        results = ext_module.dynamic_point_to_voxel_forward(
+            feats, coors, reduce_type)
+        (voxel_feats, voxel_coors, point2voxel_map,
+         voxel_points_count) = results
+        ctx.reduce_type = reduce_type
+        ctx.save_for_backward(feats, voxel_feats, point2voxel_map,
+                              voxel_points_count)
+        ctx.mark_non_differentiable(voxel_coors)
+        return voxel_feats, voxel_coors
+
+    @staticmethod
+    def backward(ctx, grad_voxel_feats, grad_voxel_coors=None):
+        (feats, voxel_feats, point2voxel_map,
+         voxel_points_count) = ctx.saved_tensors
+        grad_feats = torch.zeros_like(feats)
+        # TODO: whether to use index put or use cuda_backward
+        # To use index put, need point to voxel index
+        ext_module.dynamic_point_to_voxel_backward(
+            grad_feats, grad_voxel_feats.contiguous(), feats, voxel_feats,
+            point2voxel_map, voxel_points_count, ctx.reduce_type)
+        return grad_feats, None, None
+
+
+dynamic_scatter = _DynamicScatter.apply
+
+
+class DynamicScatter(nn.Module):
+    """Scatters points into voxels, used in the voxel encoder with dynamic
+    voxelization.
+
+    Note:
+        The CPU and GPU implementation get the same output, but have numerical
+        difference after summation and division (e.g., 5e-7).
+
+    Args:
+        voxel_size (list): list [x, y, z] size of three dimension.
+        point_cloud_range (list): The coordinate range of points, [x_min,
+            y_min, z_min, x_max, y_max, z_max].
+        average_points (bool): whether to use avg pooling to scatter points
+            into voxel.
+    """
+
+    def __init__(self, voxel_size, point_cloud_range, average_points: bool):
+        super().__init__()
+
+        self.voxel_size = voxel_size
+        self.point_cloud_range = point_cloud_range
+        self.average_points = average_points
+
+    def forward_single(self, points, coors):
+        """Scatters points into voxels.
+
+        Args:
+            points (torch.Tensor): Points to be reduced into voxels.
+            coors (torch.Tensor): Corresponding voxel coordinates (specifically
+                multi-dim voxel index) of each points.
+
+        Returns:
+            voxel_feats (torch.Tensor): Reduced features, input features that
+                shares the same voxel coordinates are reduced to one row.
+            voxel_coors (torch.Tensor): Voxel coordinates.
+        """
+        reduce = 'mean' if self.average_points else 'max'
+        return dynamic_scatter(points.contiguous(), coors.contiguous(), reduce)
+
+    def forward(self, points, coors):
+        """Scatters points/features into voxels.
+
+        Args:
+            points (torch.Tensor): Points to be reduced into voxels.
+            coors (torch.Tensor): Corresponding voxel coordinates (specifically
+                multi-dim voxel index) of each points.
+
+        Returns:
+            voxel_feats (torch.Tensor): Reduced features, input features that
+                shares the same voxel coordinates are reduced to one row.
+            voxel_coors (torch.Tensor): Voxel coordinates.
+        """
+        if coors.size(-1) == 3:
+            return self.forward_single(points, coors)
+        else:
+            batch_size = coors[-1, 0] + 1
+            voxels, voxel_coors = [], []
+            for i in range(batch_size):
+                inds = torch.where(coors[:, 0] == i)
+                voxel, voxel_coor = self.forward_single(
+                    points[inds], coors[inds][:, 1:])
+                coor_pad = nn.functional.pad(
+                    voxel_coor, (1, 0), mode='constant', value=i)
+                voxel_coors.append(coor_pad)
+                voxels.append(voxel)
+            features = torch.cat(voxels, dim=0)
+            feature_coors = torch.cat(voxel_coors, dim=0)
+
+            return features, feature_coors
+
+    def __repr__(self):
+        s = self.__class__.__name__ + '('
+        s += 'voxel_size=' + str(self.voxel_size)
+        s += ', point_cloud_range=' + str(self.point_cloud_range)
+        s += ', average_points=' + str(self.average_points)
+        s += ')'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/sync_bn.py b/annotator/uniformer/mmcv/ops/sync_bn.py
new file mode 100644
index 0000000000000000000000000000000000000000..c9b016fcbe860989c56cd1040034bcfa60e146d2
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/sync_bn.py
@@ -0,0 +1,279 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.distributed as dist
+import torch.nn.functional as F
+from torch.autograd import Function
+from torch.autograd.function import once_differentiable
+from torch.nn.modules.module import Module
+from torch.nn.parameter import Parameter
+
+from annotator.uniformer.mmcv.cnn import NORM_LAYERS
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', [
+    'sync_bn_forward_mean', 'sync_bn_forward_var', 'sync_bn_forward_output',
+    'sync_bn_backward_param', 'sync_bn_backward_data'
+])
+
+
+class SyncBatchNormFunction(Function):
+
+    @staticmethod
+    def symbolic(g, input, running_mean, running_var, weight, bias, momentum,
+                 eps, group, group_size, stats_mode):
+        return g.op(
+            'mmcv::MMCVSyncBatchNorm',
+            input,
+            running_mean,
+            running_var,
+            weight,
+            bias,
+            momentum_f=momentum,
+            eps_f=eps,
+            group_i=group,
+            group_size_i=group_size,
+            stats_mode=stats_mode)
+
+    @staticmethod
+    def forward(self, input, running_mean, running_var, weight, bias, momentum,
+                eps, group, group_size, stats_mode):
+        self.momentum = momentum
+        self.eps = eps
+        self.group = group
+        self.group_size = group_size
+        self.stats_mode = stats_mode
+
+        assert isinstance(
+                   input, (torch.HalfTensor, torch.FloatTensor,
+                           torch.cuda.HalfTensor, torch.cuda.FloatTensor)), \
+               f'only support Half or Float Tensor, but {input.type()}'
+        output = torch.zeros_like(input)
+        input3d = input.flatten(start_dim=2)
+        output3d = output.view_as(input3d)
+        num_channels = input3d.size(1)
+
+        # ensure mean/var/norm/std are initialized as zeros
+        # ``torch.empty()`` does not guarantee that
+        mean = torch.zeros(
+            num_channels, dtype=torch.float, device=input3d.device)
+        var = torch.zeros(
+            num_channels, dtype=torch.float, device=input3d.device)
+        norm = torch.zeros_like(
+            input3d, dtype=torch.float, device=input3d.device)
+        std = torch.zeros(
+            num_channels, dtype=torch.float, device=input3d.device)
+
+        batch_size = input3d.size(0)
+        if batch_size > 0:
+            ext_module.sync_bn_forward_mean(input3d, mean)
+            batch_flag = torch.ones([1], device=mean.device, dtype=mean.dtype)
+        else:
+            # skip updating mean and leave it as zeros when the input is empty
+            batch_flag = torch.zeros([1], device=mean.device, dtype=mean.dtype)
+
+        # synchronize mean and the batch flag
+        vec = torch.cat([mean, batch_flag])
+        if self.stats_mode == 'N':
+            vec *= batch_size
+        if self.group_size > 1:
+            dist.all_reduce(vec, group=self.group)
+        total_batch = vec[-1].detach()
+        mean = vec[:num_channels]
+
+        if self.stats_mode == 'default':
+            mean = mean / self.group_size
+        elif self.stats_mode == 'N':
+            mean = mean / total_batch.clamp(min=1)
+        else:
+            raise NotImplementedError
+
+        # leave var as zeros when the input is empty
+        if batch_size > 0:
+            ext_module.sync_bn_forward_var(input3d, mean, var)
+
+        if self.stats_mode == 'N':
+            var *= batch_size
+        if self.group_size > 1:
+            dist.all_reduce(var, group=self.group)
+
+        if self.stats_mode == 'default':
+            var /= self.group_size
+        elif self.stats_mode == 'N':
+            var /= total_batch.clamp(min=1)
+        else:
+            raise NotImplementedError
+
+        # if the total batch size over all the ranks is zero,
+        # we should not update the statistics in the current batch
+        update_flag = total_batch.clamp(max=1)
+        momentum = update_flag * self.momentum
+        ext_module.sync_bn_forward_output(
+            input3d,
+            mean,
+            var,
+            weight,
+            bias,
+            running_mean,
+            running_var,
+            norm,
+            std,
+            output3d,
+            eps=self.eps,
+            momentum=momentum,
+            group_size=self.group_size)
+        self.save_for_backward(norm, std, weight)
+        return output
+
+    @staticmethod
+    @once_differentiable
+    def backward(self, grad_output):
+        norm, std, weight = self.saved_tensors
+        grad_weight = torch.zeros_like(weight)
+        grad_bias = torch.zeros_like(weight)
+        grad_input = torch.zeros_like(grad_output)
+        grad_output3d = grad_output.flatten(start_dim=2)
+        grad_input3d = grad_input.view_as(grad_output3d)
+
+        batch_size = grad_input3d.size(0)
+        if batch_size > 0:
+            ext_module.sync_bn_backward_param(grad_output3d, norm, grad_weight,
+                                              grad_bias)
+
+        # all reduce
+        if self.group_size > 1:
+            dist.all_reduce(grad_weight, group=self.group)
+            dist.all_reduce(grad_bias, group=self.group)
+            grad_weight /= self.group_size
+            grad_bias /= self.group_size
+
+        if batch_size > 0:
+            ext_module.sync_bn_backward_data(grad_output3d, weight,
+                                             grad_weight, grad_bias, norm, std,
+                                             grad_input3d)
+
+        return grad_input, None, None, grad_weight, grad_bias, \
+            None, None, None, None, None
+
+
+@NORM_LAYERS.register_module(name='MMSyncBN')
+class SyncBatchNorm(Module):
+    """Synchronized Batch Normalization.
+
+    Args:
+        num_features (int): number of features/chennels in input tensor
+        eps (float, optional): a value added to the denominator for numerical
+            stability. Defaults to 1e-5.
+        momentum (float, optional): the value used for the running_mean and
+            running_var computation. Defaults to 0.1.
+        affine (bool, optional): whether to use learnable affine parameters.
+            Defaults to True.
+        track_running_stats (bool, optional): whether to track the running
+            mean and variance during training. When set to False, this
+            module does not track such statistics, and initializes statistics
+            buffers ``running_mean`` and ``running_var`` as ``None``. When
+            these buffers are ``None``, this module always uses batch
+            statistics in both training and eval modes. Defaults to True.
+        group (int, optional): synchronization of stats happen within
+            each process group individually. By default it is synchronization
+            across the whole world. Defaults to None.
+        stats_mode (str, optional): The statistical mode. Available options
+            includes ``'default'`` and ``'N'``. Defaults to 'default'.
+            When ``stats_mode=='default'``, it computes the overall statistics
+            using those from each worker with equal weight, i.e., the
+            statistics are synchronized and simply divied by ``group``. This
+            mode will produce inaccurate statistics when empty tensors occur.
+            When ``stats_mode=='N'``, it compute the overall statistics using
+            the total number of batches in each worker ignoring the number of
+            group, i.e., the statistics are synchronized and then divied by
+            the total batch ``N``. This mode is beneficial when empty tensors
+            occur during training, as it average the total mean by the real
+            number of batch.
+    """
+
+    def __init__(self,
+                 num_features,
+                 eps=1e-5,
+                 momentum=0.1,
+                 affine=True,
+                 track_running_stats=True,
+                 group=None,
+                 stats_mode='default'):
+        super(SyncBatchNorm, self).__init__()
+        self.num_features = num_features
+        self.eps = eps
+        self.momentum = momentum
+        self.affine = affine
+        self.track_running_stats = track_running_stats
+        group = dist.group.WORLD if group is None else group
+        self.group = group
+        self.group_size = dist.get_world_size(group)
+        assert stats_mode in ['default', 'N'], \
+            f'"stats_mode" only accepts "default" and "N", got "{stats_mode}"'
+        self.stats_mode = stats_mode
+        if self.affine:
+            self.weight = Parameter(torch.Tensor(num_features))
+            self.bias = Parameter(torch.Tensor(num_features))
+        else:
+            self.register_parameter('weight', None)
+            self.register_parameter('bias', None)
+        if self.track_running_stats:
+            self.register_buffer('running_mean', torch.zeros(num_features))
+            self.register_buffer('running_var', torch.ones(num_features))
+            self.register_buffer('num_batches_tracked',
+                                 torch.tensor(0, dtype=torch.long))
+        else:
+            self.register_buffer('running_mean', None)
+            self.register_buffer('running_var', None)
+            self.register_buffer('num_batches_tracked', None)
+        self.reset_parameters()
+
+    def reset_running_stats(self):
+        if self.track_running_stats:
+            self.running_mean.zero_()
+            self.running_var.fill_(1)
+            self.num_batches_tracked.zero_()
+
+    def reset_parameters(self):
+        self.reset_running_stats()
+        if self.affine:
+            self.weight.data.uniform_()  # pytorch use ones_()
+            self.bias.data.zero_()
+
+    def forward(self, input):
+        if input.dim() < 2:
+            raise ValueError(
+                f'expected at least 2D input, got {input.dim()}D input')
+        if self.momentum is None:
+            exponential_average_factor = 0.0
+        else:
+            exponential_average_factor = self.momentum
+
+        if self.training and self.track_running_stats:
+            if self.num_batches_tracked is not None:
+                self.num_batches_tracked += 1
+                if self.momentum is None:  # use cumulative moving average
+                    exponential_average_factor = 1.0 / float(
+                        self.num_batches_tracked)
+                else:  # use exponential moving average
+                    exponential_average_factor = self.momentum
+
+        if self.training or not self.track_running_stats:
+            return SyncBatchNormFunction.apply(
+                input, self.running_mean, self.running_var, self.weight,
+                self.bias, exponential_average_factor, self.eps, self.group,
+                self.group_size, self.stats_mode)
+        else:
+            return F.batch_norm(input, self.running_mean, self.running_var,
+                                self.weight, self.bias, False,
+                                exponential_average_factor, self.eps)
+
+    def __repr__(self):
+        s = self.__class__.__name__
+        s += f'({self.num_features}, '
+        s += f'eps={self.eps}, '
+        s += f'momentum={self.momentum}, '
+        s += f'affine={self.affine}, '
+        s += f'track_running_stats={self.track_running_stats}, '
+        s += f'group_size={self.group_size},'
+        s += f'stats_mode={self.stats_mode})'
+        return s
diff --git a/annotator/uniformer/mmcv/ops/three_interpolate.py b/annotator/uniformer/mmcv/ops/three_interpolate.py
new file mode 100644
index 0000000000000000000000000000000000000000..203f47f05d58087e034fb3cd8cd6a09233947b4a
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/three_interpolate.py
@@ -0,0 +1,68 @@
+from typing import Tuple
+
+import torch
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['three_interpolate_forward', 'three_interpolate_backward'])
+
+
+class ThreeInterpolate(Function):
+    """Performs weighted linear interpolation on 3 features.
+
+    Please refer to `Paper of PointNet++ <https://arxiv.org/abs/1706.02413>`_
+    for more details.
+    """
+
+    @staticmethod
+    def forward(ctx, features: torch.Tensor, indices: torch.Tensor,
+                weight: torch.Tensor) -> torch.Tensor:
+        """
+        Args:
+            features (Tensor): (B, C, M) Features descriptors to be
+                interpolated
+            indices (Tensor): (B, n, 3) index three nearest neighbors
+                of the target features in features
+            weight (Tensor): (B, n, 3) weights of interpolation
+
+        Returns:
+            Tensor: (B, C, N) tensor of the interpolated features
+        """
+        assert features.is_contiguous()
+        assert indices.is_contiguous()
+        assert weight.is_contiguous()
+
+        B, c, m = features.size()
+        n = indices.size(1)
+        ctx.three_interpolate_for_backward = (indices, weight, m)
+        output = torch.cuda.FloatTensor(B, c, n)
+
+        ext_module.three_interpolate_forward(
+            features, indices, weight, output, b=B, c=c, m=m, n=n)
+        return output
+
+    @staticmethod
+    def backward(
+        ctx, grad_out: torch.Tensor
+    ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
+        """
+        Args:
+            grad_out (Tensor): (B, C, N) tensor with gradients of outputs
+
+        Returns:
+            Tensor: (B, C, M) tensor with gradients of features
+        """
+        idx, weight, m = ctx.three_interpolate_for_backward
+        B, c, n = grad_out.size()
+
+        grad_features = torch.cuda.FloatTensor(B, c, m).zero_()
+        grad_out_data = grad_out.data.contiguous()
+
+        ext_module.three_interpolate_backward(
+            grad_out_data, idx, weight, grad_features.data, b=B, c=c, n=n, m=m)
+        return grad_features, None, None
+
+
+three_interpolate = ThreeInterpolate.apply
diff --git a/annotator/uniformer/mmcv/ops/three_nn.py b/annotator/uniformer/mmcv/ops/three_nn.py
new file mode 100644
index 0000000000000000000000000000000000000000..2b01047a129989cd5545a0a86f23a487f4a13ce1
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/three_nn.py
@@ -0,0 +1,51 @@
+from typing import Tuple
+
+import torch
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext', ['three_nn_forward'])
+
+
+class ThreeNN(Function):
+    """Find the top-3 nearest neighbors of the target set from the source set.
+
+    Please refer to `Paper of PointNet++ <https://arxiv.org/abs/1706.02413>`_
+    for more details.
+    """
+
+    @staticmethod
+    def forward(ctx, target: torch.Tensor,
+                source: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        """
+        Args:
+            target (Tensor): shape (B, N, 3), points set that needs to
+                find the nearest neighbors.
+            source (Tensor): shape (B, M, 3), points set that is used
+                to find the nearest neighbors of points in target set.
+
+        Returns:
+            Tensor: shape (B, N, 3), L2 distance of each point in target
+                set to their corresponding nearest neighbors.
+        """
+        target = target.contiguous()
+        source = source.contiguous()
+
+        B, N, _ = target.size()
+        m = source.size(1)
+        dist2 = torch.cuda.FloatTensor(B, N, 3)
+        idx = torch.cuda.IntTensor(B, N, 3)
+
+        ext_module.three_nn_forward(target, source, dist2, idx, b=B, n=N, m=m)
+        if torch.__version__ != 'parrots':
+            ctx.mark_non_differentiable(idx)
+
+        return torch.sqrt(dist2), idx
+
+    @staticmethod
+    def backward(ctx, a=None, b=None):
+        return None, None
+
+
+three_nn = ThreeNN.apply
diff --git a/annotator/uniformer/mmcv/ops/tin_shift.py b/annotator/uniformer/mmcv/ops/tin_shift.py
new file mode 100644
index 0000000000000000000000000000000000000000..472c9fcfe45a124e819b7ed5653e585f94a8811e
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/tin_shift.py
@@ -0,0 +1,68 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+# Code reference from "Temporal Interlacing Network"
+# https://github.com/deepcs233/TIN/blob/master/cuda_shift/rtc_wrap.py
+# Hao Shao, Shengju Qian, Yu Liu
+# shaoh19@mails.tsinghua.edu.cn, sjqian@cse.cuhk.edu.hk, yuliu@ee.cuhk.edu.hk
+
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext('_ext',
+                                 ['tin_shift_forward', 'tin_shift_backward'])
+
+
+class TINShiftFunction(Function):
+
+    @staticmethod
+    def forward(ctx, input, shift):
+        C = input.size(2)
+        num_segments = shift.size(1)
+        if C // num_segments <= 0 or C % num_segments != 0:
+            raise ValueError('C should be a multiple of num_segments, '
+                             f'but got C={C} and num_segments={num_segments}.')
+
+        ctx.save_for_backward(shift)
+
+        out = torch.zeros_like(input)
+        ext_module.tin_shift_forward(input, shift, out)
+
+        return out
+
+    @staticmethod
+    def backward(ctx, grad_output):
+
+        shift = ctx.saved_tensors[0]
+        data_grad_input = grad_output.new(*grad_output.size()).zero_()
+        shift_grad_input = shift.new(*shift.size()).zero_()
+        ext_module.tin_shift_backward(grad_output, shift, data_grad_input)
+
+        return data_grad_input, shift_grad_input
+
+
+tin_shift = TINShiftFunction.apply
+
+
+class TINShift(nn.Module):
+    """Temporal Interlace Shift.
+
+    Temporal Interlace shift is a differentiable temporal-wise frame shifting
+    which is proposed in "Temporal Interlacing Network"
+
+    Please refer to https://arxiv.org/abs/2001.06499 for more details.
+    Code is modified from https://github.com/mit-han-lab/temporal-shift-module
+    """
+
+    def forward(self, input, shift):
+        """Perform temporal interlace shift.
+
+        Args:
+            input (Tensor): Feature map with shape [N, num_segments, C, H * W].
+            shift (Tensor): Shift tensor with shape [N, num_segments].
+
+        Returns:
+            Feature map after temporal interlace shift.
+        """
+        return tin_shift(input, shift)
diff --git a/annotator/uniformer/mmcv/ops/upfirdn2d.py b/annotator/uniformer/mmcv/ops/upfirdn2d.py
new file mode 100644
index 0000000000000000000000000000000000000000..c8bb2c3c949eed38a6465ed369fa881538dca010
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/upfirdn2d.py
@@ -0,0 +1,330 @@
+# modified from https://github.com/rosinality/stylegan2-pytorch/blob/master/op/upfirdn2d.py  # noqa:E501
+
+# Copyright (c) 2021, NVIDIA Corporation. All rights reserved.
+# NVIDIA Source Code License for StyleGAN2 with Adaptive Discriminator
+# Augmentation (ADA)
+# =======================================================================
+
+# 1. Definitions
+
+# "Licensor" means any person or entity that distributes its Work.
+
+# "Software" means the original work of authorship made available under
+# this License.
+
+# "Work" means the Software and any additions to or derivative works of
+# the Software that are made available under this License.
+
+# The terms "reproduce," "reproduction," "derivative works," and
+# "distribution" have the meaning as provided under U.S. copyright law;
+# provided, however, that for the purposes of this License, derivative
+# works shall not include works that remain separable from, or merely
+# link (or bind by name) to the interfaces of, the Work.
+
+# Works, including the Software, are "made available" under this License
+# by including in or with the Work either (a) a copyright notice
+# referencing the applicability of this License to the Work, or (b) a
+# copy of this License.
+
+# 2. License Grants
+
+#     2.1 Copyright Grant. Subject to the terms and conditions of this
+#     License, each Licensor grants to you a perpetual, worldwide,
+#     non-exclusive, royalty-free, copyright license to reproduce,
+#     prepare derivative works of, publicly display, publicly perform,
+#     sublicense and distribute its Work and any resulting derivative
+#     works in any form.
+
+# 3. Limitations
+
+#     3.1 Redistribution. You may reproduce or distribute the Work only
+#     if (a) you do so under this License, (b) you include a complete
+#     copy of this License with your distribution, and (c) you retain
+#     without modification any copyright, patent, trademark, or
+#     attribution notices that are present in the Work.
+
+#     3.2 Derivative Works. You may specify that additional or different
+#     terms apply to the use, reproduction, and distribution of your
+#     derivative works of the Work ("Your Terms") only if (a) Your Terms
+#     provide that the use limitation in Section 3.3 applies to your
+#     derivative works, and (b) you identify the specific derivative
+#     works that are subject to Your Terms. Notwithstanding Your Terms,
+#     this License (including the redistribution requirements in Section
+#     3.1) will continue to apply to the Work itself.
+
+#     3.3 Use Limitation. The Work and any derivative works thereof only
+#     may be used or intended for use non-commercially. Notwithstanding
+#     the foregoing, NVIDIA and its affiliates may use the Work and any
+#     derivative works commercially. As used herein, "non-commercially"
+#     means for research or evaluation purposes only.
+
+#     3.4 Patent Claims. If you bring or threaten to bring a patent claim
+#     against any Licensor (including any claim, cross-claim or
+#     counterclaim in a lawsuit) to enforce any patents that you allege
+#     are infringed by any Work, then your rights under this License from
+#     such Licensor (including the grant in Section 2.1) will terminate
+#     immediately.
+
+#     3.5 Trademarks. This License does not grant any rights to use any
+#     Licensor’s or its affiliates’ names, logos, or trademarks, except
+#     as necessary to reproduce the notices described in this License.
+
+#     3.6 Termination. If you violate any term of this License, then your
+#     rights under this License (including the grant in Section 2.1) will
+#     terminate immediately.
+
+# 4. Disclaimer of Warranty.
+
+# THE WORK IS PROVIDED "AS IS" WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WARRANTIES OR CONDITIONS OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE OR
+# NON-INFRINGEMENT. YOU BEAR THE RISK OF UNDERTAKING ANY ACTIVITIES UNDER
+# THIS LICENSE.
+
+# 5. Limitation of Liability.
+
+# EXCEPT AS PROHIBITED BY APPLICABLE LAW, IN NO EVENT AND UNDER NO LEGAL
+# THEORY, WHETHER IN TORT (INCLUDING NEGLIGENCE), CONTRACT, OR OTHERWISE
+# SHALL ANY LICENSOR BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY DIRECT,
+# INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF
+# OR RELATED TO THIS LICENSE, THE USE OR INABILITY TO USE THE WORK
+# (INCLUDING BUT NOT LIMITED TO LOSS OF GOODWILL, BUSINESS INTERRUPTION,
+# LOST PROFITS OR DATA, COMPUTER FAILURE OR MALFUNCTION, OR ANY OTHER
+# COMMERCIAL DAMAGES OR LOSSES), EVEN IF THE LICENSOR HAS BEEN ADVISED OF
+# THE POSSIBILITY OF SUCH DAMAGES.
+
+# =======================================================================
+
+import torch
+from torch.autograd import Function
+from torch.nn import functional as F
+
+from annotator.uniformer.mmcv.utils import to_2tuple
+from ..utils import ext_loader
+
+upfirdn2d_ext = ext_loader.load_ext('_ext', ['upfirdn2d'])
+
+
+class UpFirDn2dBackward(Function):
+
+    @staticmethod
+    def forward(ctx, grad_output, kernel, grad_kernel, up, down, pad, g_pad,
+                in_size, out_size):
+
+        up_x, up_y = up
+        down_x, down_y = down
+        g_pad_x0, g_pad_x1, g_pad_y0, g_pad_y1 = g_pad
+
+        grad_output = grad_output.reshape(-1, out_size[0], out_size[1], 1)
+
+        grad_input = upfirdn2d_ext.upfirdn2d(
+            grad_output,
+            grad_kernel,
+            up_x=down_x,
+            up_y=down_y,
+            down_x=up_x,
+            down_y=up_y,
+            pad_x0=g_pad_x0,
+            pad_x1=g_pad_x1,
+            pad_y0=g_pad_y0,
+            pad_y1=g_pad_y1)
+        grad_input = grad_input.view(in_size[0], in_size[1], in_size[2],
+                                     in_size[3])
+
+        ctx.save_for_backward(kernel)
+
+        pad_x0, pad_x1, pad_y0, pad_y1 = pad
+
+        ctx.up_x = up_x
+        ctx.up_y = up_y
+        ctx.down_x = down_x
+        ctx.down_y = down_y
+        ctx.pad_x0 = pad_x0
+        ctx.pad_x1 = pad_x1
+        ctx.pad_y0 = pad_y0
+        ctx.pad_y1 = pad_y1
+        ctx.in_size = in_size
+        ctx.out_size = out_size
+
+        return grad_input
+
+    @staticmethod
+    def backward(ctx, gradgrad_input):
+        kernel, = ctx.saved_tensors
+
+        gradgrad_input = gradgrad_input.reshape(-1, ctx.in_size[2],
+                                                ctx.in_size[3], 1)
+
+        gradgrad_out = upfirdn2d_ext.upfirdn2d(
+            gradgrad_input,
+            kernel,
+            up_x=ctx.up_x,
+            up_y=ctx.up_y,
+            down_x=ctx.down_x,
+            down_y=ctx.down_y,
+            pad_x0=ctx.pad_x0,
+            pad_x1=ctx.pad_x1,
+            pad_y0=ctx.pad_y0,
+            pad_y1=ctx.pad_y1)
+        # gradgrad_out = gradgrad_out.view(ctx.in_size[0], ctx.out_size[0],
+        #                                  ctx.out_size[1], ctx.in_size[3])
+        gradgrad_out = gradgrad_out.view(ctx.in_size[0], ctx.in_size[1],
+                                         ctx.out_size[0], ctx.out_size[1])
+
+        return gradgrad_out, None, None, None, None, None, None, None, None
+
+
+class UpFirDn2d(Function):
+
+    @staticmethod
+    def forward(ctx, input, kernel, up, down, pad):
+        up_x, up_y = up
+        down_x, down_y = down
+        pad_x0, pad_x1, pad_y0, pad_y1 = pad
+
+        kernel_h, kernel_w = kernel.shape
+        batch, channel, in_h, in_w = input.shape
+        ctx.in_size = input.shape
+
+        input = input.reshape(-1, in_h, in_w, 1)
+
+        ctx.save_for_backward(kernel, torch.flip(kernel, [0, 1]))
+
+        out_h = (in_h * up_y + pad_y0 + pad_y1 - kernel_h) // down_y + 1
+        out_w = (in_w * up_x + pad_x0 + pad_x1 - kernel_w) // down_x + 1
+        ctx.out_size = (out_h, out_w)
+
+        ctx.up = (up_x, up_y)
+        ctx.down = (down_x, down_y)
+        ctx.pad = (pad_x0, pad_x1, pad_y0, pad_y1)
+
+        g_pad_x0 = kernel_w - pad_x0 - 1
+        g_pad_y0 = kernel_h - pad_y0 - 1
+        g_pad_x1 = in_w * up_x - out_w * down_x + pad_x0 - up_x + 1
+        g_pad_y1 = in_h * up_y - out_h * down_y + pad_y0 - up_y + 1
+
+        ctx.g_pad = (g_pad_x0, g_pad_x1, g_pad_y0, g_pad_y1)
+
+        out = upfirdn2d_ext.upfirdn2d(
+            input,
+            kernel,
+            up_x=up_x,
+            up_y=up_y,
+            down_x=down_x,
+            down_y=down_y,
+            pad_x0=pad_x0,
+            pad_x1=pad_x1,
+            pad_y0=pad_y0,
+            pad_y1=pad_y1)
+        # out = out.view(major, out_h, out_w, minor)
+        out = out.view(-1, channel, out_h, out_w)
+
+        return out
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        kernel, grad_kernel = ctx.saved_tensors
+
+        grad_input = UpFirDn2dBackward.apply(
+            grad_output,
+            kernel,
+            grad_kernel,
+            ctx.up,
+            ctx.down,
+            ctx.pad,
+            ctx.g_pad,
+            ctx.in_size,
+            ctx.out_size,
+        )
+
+        return grad_input, None, None, None, None
+
+
+def upfirdn2d(input, kernel, up=1, down=1, pad=(0, 0)):
+    """UpFRIDn for 2d features.
+
+    UpFIRDn is short for upsample, apply FIR filter and downsample. More
+    details can be found in:
+    https://www.mathworks.com/help/signal/ref/upfirdn.html
+
+    Args:
+        input (Tensor): Tensor with shape of (n, c, h, w).
+        kernel (Tensor): Filter kernel.
+        up (int | tuple[int], optional): Upsampling factor. If given a number,
+            we will use this factor for the both height and width side.
+            Defaults to 1.
+        down (int | tuple[int], optional): Downsampling factor. If given a
+            number, we will use this factor for the both height and width side.
+            Defaults to 1.
+        pad (tuple[int], optional): Padding for tensors, (x_pad, y_pad) or
+            (x_pad_0, x_pad_1, y_pad_0, y_pad_1). Defaults to (0, 0).
+
+    Returns:
+        Tensor: Tensor after UpFIRDn.
+    """
+    if input.device.type == 'cpu':
+        if len(pad) == 2:
+            pad = (pad[0], pad[1], pad[0], pad[1])
+
+        up = to_2tuple(up)
+
+        down = to_2tuple(down)
+
+        out = upfirdn2d_native(input, kernel, up[0], up[1], down[0], down[1],
+                               pad[0], pad[1], pad[2], pad[3])
+    else:
+        _up = to_2tuple(up)
+
+        _down = to_2tuple(down)
+
+        if len(pad) == 4:
+            _pad = pad
+        elif len(pad) == 2:
+            _pad = (pad[0], pad[1], pad[0], pad[1])
+
+        out = UpFirDn2d.apply(input, kernel, _up, _down, _pad)
+
+    return out
+
+
+def upfirdn2d_native(input, kernel, up_x, up_y, down_x, down_y, pad_x0, pad_x1,
+                     pad_y0, pad_y1):
+    _, channel, in_h, in_w = input.shape
+    input = input.reshape(-1, in_h, in_w, 1)
+
+    _, in_h, in_w, minor = input.shape
+    kernel_h, kernel_w = kernel.shape
+
+    out = input.view(-1, in_h, 1, in_w, 1, minor)
+    out = F.pad(out, [0, 0, 0, up_x - 1, 0, 0, 0, up_y - 1])
+    out = out.view(-1, in_h * up_y, in_w * up_x, minor)
+
+    out = F.pad(
+        out,
+        [0, 0,
+         max(pad_x0, 0),
+         max(pad_x1, 0),
+         max(pad_y0, 0),
+         max(pad_y1, 0)])
+    out = out[:,
+              max(-pad_y0, 0):out.shape[1] - max(-pad_y1, 0),
+              max(-pad_x0, 0):out.shape[2] - max(-pad_x1, 0), :, ]
+
+    out = out.permute(0, 3, 1, 2)
+    out = out.reshape(
+        [-1, 1, in_h * up_y + pad_y0 + pad_y1, in_w * up_x + pad_x0 + pad_x1])
+    w = torch.flip(kernel, [0, 1]).view(1, 1, kernel_h, kernel_w)
+    out = F.conv2d(out, w)
+    out = out.reshape(
+        -1,
+        minor,
+        in_h * up_y + pad_y0 + pad_y1 - kernel_h + 1,
+        in_w * up_x + pad_x0 + pad_x1 - kernel_w + 1,
+    )
+    out = out.permute(0, 2, 3, 1)
+    out = out[:, ::down_y, ::down_x, :]
+
+    out_h = (in_h * up_y + pad_y0 + pad_y1 - kernel_h) // down_y + 1
+    out_w = (in_w * up_x + pad_x0 + pad_x1 - kernel_w) // down_x + 1
+
+    return out.view(-1, channel, out_h, out_w)
diff --git a/annotator/uniformer/mmcv/ops/voxelize.py b/annotator/uniformer/mmcv/ops/voxelize.py
new file mode 100644
index 0000000000000000000000000000000000000000..ca3226a4fbcbfe58490fa2ea8e1c16b531214121
--- /dev/null
+++ b/annotator/uniformer/mmcv/ops/voxelize.py
@@ -0,0 +1,132 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch import nn
+from torch.autograd import Function
+from torch.nn.modules.utils import _pair
+
+from ..utils import ext_loader
+
+ext_module = ext_loader.load_ext(
+    '_ext', ['dynamic_voxelize_forward', 'hard_voxelize_forward'])
+
+
+class _Voxelization(Function):
+
+    @staticmethod
+    def forward(ctx,
+                points,
+                voxel_size,
+                coors_range,
+                max_points=35,
+                max_voxels=20000):
+        """Convert kitti points(N, >=3) to voxels.
+
+        Args:
+            points (torch.Tensor): [N, ndim]. Points[:, :3] contain xyz points
+                and points[:, 3:] contain other information like reflectivity.
+            voxel_size (tuple or float): The size of voxel with the shape of
+                [3].
+            coors_range (tuple or float): The coordinate range of voxel with
+                the shape of [6].
+            max_points (int, optional): maximum points contained in a voxel. if
+                max_points=-1, it means using dynamic_voxelize. Default: 35.
+            max_voxels (int, optional): maximum voxels this function create.
+                for second, 20000 is a good choice. Users should shuffle points
+                before call this function because max_voxels may drop points.
+                Default: 20000.
+
+        Returns:
+            voxels_out (torch.Tensor): Output voxels with the shape of [M,
+                max_points, ndim]. Only contain points and returned when
+                max_points != -1.
+            coors_out (torch.Tensor): Output coordinates with the shape of
+                [M, 3].
+            num_points_per_voxel_out (torch.Tensor): Num points per voxel with
+                the shape of [M]. Only returned when max_points != -1.
+        """
+        if max_points == -1 or max_voxels == -1:
+            coors = points.new_zeros(size=(points.size(0), 3), dtype=torch.int)
+            ext_module.dynamic_voxelize_forward(points, coors, voxel_size,
+                                                coors_range, 3)
+            return coors
+        else:
+            voxels = points.new_zeros(
+                size=(max_voxels, max_points, points.size(1)))
+            coors = points.new_zeros(size=(max_voxels, 3), dtype=torch.int)
+            num_points_per_voxel = points.new_zeros(
+                size=(max_voxels, ), dtype=torch.int)
+            voxel_num = ext_module.hard_voxelize_forward(
+                points, voxels, coors, num_points_per_voxel, voxel_size,
+                coors_range, max_points, max_voxels, 3)
+            # select the valid voxels
+            voxels_out = voxels[:voxel_num]
+            coors_out = coors[:voxel_num]
+            num_points_per_voxel_out = num_points_per_voxel[:voxel_num]
+            return voxels_out, coors_out, num_points_per_voxel_out
+
+
+voxelization = _Voxelization.apply
+
+
+class Voxelization(nn.Module):
+    """Convert kitti points(N, >=3) to voxels.
+
+    Please refer to `PVCNN <https://arxiv.org/abs/1907.03739>`_ for more
+    details.
+
+    Args:
+        voxel_size (tuple or float): The size of voxel with the shape of [3].
+        point_cloud_range (tuple or float): The coordinate range of voxel with
+            the shape of [6].
+        max_num_points (int): maximum points contained in a voxel. if
+            max_points=-1, it means using dynamic_voxelize.
+        max_voxels (int, optional): maximum voxels this function create.
+            for second, 20000 is a good choice. Users should shuffle points
+            before call this function because max_voxels may drop points.
+            Default: 20000.
+    """
+
+    def __init__(self,
+                 voxel_size,
+                 point_cloud_range,
+                 max_num_points,
+                 max_voxels=20000):
+        super().__init__()
+
+        self.voxel_size = voxel_size
+        self.point_cloud_range = point_cloud_range
+        self.max_num_points = max_num_points
+        if isinstance(max_voxels, tuple):
+            self.max_voxels = max_voxels
+        else:
+            self.max_voxels = _pair(max_voxels)
+
+        point_cloud_range = torch.tensor(
+            point_cloud_range, dtype=torch.float32)
+        voxel_size = torch.tensor(voxel_size, dtype=torch.float32)
+        grid_size = (point_cloud_range[3:] -
+                     point_cloud_range[:3]) / voxel_size
+        grid_size = torch.round(grid_size).long()
+        input_feat_shape = grid_size[:2]
+        self.grid_size = grid_size
+        # the origin shape is as [x-len, y-len, z-len]
+        # [w, h, d] -> [d, h, w]
+        self.pcd_shape = [*input_feat_shape, 1][::-1]
+
+    def forward(self, input):
+        if self.training:
+            max_voxels = self.max_voxels[0]
+        else:
+            max_voxels = self.max_voxels[1]
+
+        return voxelization(input, self.voxel_size, self.point_cloud_range,
+                            self.max_num_points, max_voxels)
+
+    def __repr__(self):
+        s = self.__class__.__name__ + '('
+        s += 'voxel_size=' + str(self.voxel_size)
+        s += ', point_cloud_range=' + str(self.point_cloud_range)
+        s += ', max_num_points=' + str(self.max_num_points)
+        s += ', max_voxels=' + str(self.max_voxels)
+        s += ')'
+        return s
diff --git a/annotator/uniformer/mmcv/parallel/__init__.py b/annotator/uniformer/mmcv/parallel/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..2ed2c17ad357742e423beeaf4d35db03fe9af469
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/__init__.py
@@ -0,0 +1,13 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .collate import collate
+from .data_container import DataContainer
+from .data_parallel import MMDataParallel
+from .distributed import MMDistributedDataParallel
+from .registry import MODULE_WRAPPERS
+from .scatter_gather import scatter, scatter_kwargs
+from .utils import is_module_wrapper
+
+__all__ = [
+    'collate', 'DataContainer', 'MMDataParallel', 'MMDistributedDataParallel',
+    'scatter', 'scatter_kwargs', 'is_module_wrapper', 'MODULE_WRAPPERS'
+]
diff --git a/annotator/uniformer/mmcv/parallel/_functions.py b/annotator/uniformer/mmcv/parallel/_functions.py
new file mode 100644
index 0000000000000000000000000000000000000000..9b5a8a44483ab991411d07122b22a1d027e4be8e
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/_functions.py
@@ -0,0 +1,79 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch.nn.parallel._functions import _get_stream
+
+
+def scatter(input, devices, streams=None):
+    """Scatters tensor across multiple GPUs."""
+    if streams is None:
+        streams = [None] * len(devices)
+
+    if isinstance(input, list):
+        chunk_size = (len(input) - 1) // len(devices) + 1
+        outputs = [
+            scatter(input[i], [devices[i // chunk_size]],
+                    [streams[i // chunk_size]]) for i in range(len(input))
+        ]
+        return outputs
+    elif isinstance(input, torch.Tensor):
+        output = input.contiguous()
+        # TODO: copy to a pinned buffer first (if copying from CPU)
+        stream = streams[0] if output.numel() > 0 else None
+        if devices != [-1]:
+            with torch.cuda.device(devices[0]), torch.cuda.stream(stream):
+                output = output.cuda(devices[0], non_blocking=True)
+        else:
+            # unsqueeze the first dimension thus the tensor's shape is the
+            # same as those scattered with GPU.
+            output = output.unsqueeze(0)
+        return output
+    else:
+        raise Exception(f'Unknown type {type(input)}.')
+
+
+def synchronize_stream(output, devices, streams):
+    if isinstance(output, list):
+        chunk_size = len(output) // len(devices)
+        for i in range(len(devices)):
+            for j in range(chunk_size):
+                synchronize_stream(output[i * chunk_size + j], [devices[i]],
+                                   [streams[i]])
+    elif isinstance(output, torch.Tensor):
+        if output.numel() != 0:
+            with torch.cuda.device(devices[0]):
+                main_stream = torch.cuda.current_stream()
+                main_stream.wait_stream(streams[0])
+                output.record_stream(main_stream)
+    else:
+        raise Exception(f'Unknown type {type(output)}.')
+
+
+def get_input_device(input):
+    if isinstance(input, list):
+        for item in input:
+            input_device = get_input_device(item)
+            if input_device != -1:
+                return input_device
+        return -1
+    elif isinstance(input, torch.Tensor):
+        return input.get_device() if input.is_cuda else -1
+    else:
+        raise Exception(f'Unknown type {type(input)}.')
+
+
+class Scatter:
+
+    @staticmethod
+    def forward(target_gpus, input):
+        input_device = get_input_device(input)
+        streams = None
+        if input_device == -1 and target_gpus != [-1]:
+            # Perform CPU to GPU copies in a background stream
+            streams = [_get_stream(device) for device in target_gpus]
+
+        outputs = scatter(input, target_gpus, streams)
+        # Synchronize with the copy stream
+        if streams is not None:
+            synchronize_stream(outputs, target_gpus, streams)
+
+        return tuple(outputs)
diff --git a/annotator/uniformer/mmcv/parallel/collate.py b/annotator/uniformer/mmcv/parallel/collate.py
new file mode 100644
index 0000000000000000000000000000000000000000..ad749197df21b0d74297548be5f66a696adebf7f
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/collate.py
@@ -0,0 +1,84 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from collections.abc import Mapping, Sequence
+
+import torch
+import torch.nn.functional as F
+from torch.utils.data.dataloader import default_collate
+
+from .data_container import DataContainer
+
+
+def collate(batch, samples_per_gpu=1):
+    """Puts each data field into a tensor/DataContainer with outer dimension
+    batch size.
+
+    Extend default_collate to add support for
+    :type:`~mmcv.parallel.DataContainer`. There are 3 cases.
+
+    1. cpu_only = True, e.g., meta data
+    2. cpu_only = False, stack = True, e.g., images tensors
+    3. cpu_only = False, stack = False, e.g., gt bboxes
+    """
+
+    if not isinstance(batch, Sequence):
+        raise TypeError(f'{batch.dtype} is not supported.')
+
+    if isinstance(batch[0], DataContainer):
+        stacked = []
+        if batch[0].cpu_only:
+            for i in range(0, len(batch), samples_per_gpu):
+                stacked.append(
+                    [sample.data for sample in batch[i:i + samples_per_gpu]])
+            return DataContainer(
+                stacked, batch[0].stack, batch[0].padding_value, cpu_only=True)
+        elif batch[0].stack:
+            for i in range(0, len(batch), samples_per_gpu):
+                assert isinstance(batch[i].data, torch.Tensor)
+
+                if batch[i].pad_dims is not None:
+                    ndim = batch[i].dim()
+                    assert ndim > batch[i].pad_dims
+                    max_shape = [0 for _ in range(batch[i].pad_dims)]
+                    for dim in range(1, batch[i].pad_dims + 1):
+                        max_shape[dim - 1] = batch[i].size(-dim)
+                    for sample in batch[i:i + samples_per_gpu]:
+                        for dim in range(0, ndim - batch[i].pad_dims):
+                            assert batch[i].size(dim) == sample.size(dim)
+                        for dim in range(1, batch[i].pad_dims + 1):
+                            max_shape[dim - 1] = max(max_shape[dim - 1],
+                                                     sample.size(-dim))
+                    padded_samples = []
+                    for sample in batch[i:i + samples_per_gpu]:
+                        pad = [0 for _ in range(batch[i].pad_dims * 2)]
+                        for dim in range(1, batch[i].pad_dims + 1):
+                            pad[2 * dim -
+                                1] = max_shape[dim - 1] - sample.size(-dim)
+                        padded_samples.append(
+                            F.pad(
+                                sample.data, pad, value=sample.padding_value))
+                    stacked.append(default_collate(padded_samples))
+                elif batch[i].pad_dims is None:
+                    stacked.append(
+                        default_collate([
+                            sample.data
+                            for sample in batch[i:i + samples_per_gpu]
+                        ]))
+                else:
+                    raise ValueError(
+                        'pad_dims should be either None or integers (1-3)')
+
+        else:
+            for i in range(0, len(batch), samples_per_gpu):
+                stacked.append(
+                    [sample.data for sample in batch[i:i + samples_per_gpu]])
+        return DataContainer(stacked, batch[0].stack, batch[0].padding_value)
+    elif isinstance(batch[0], Sequence):
+        transposed = zip(*batch)
+        return [collate(samples, samples_per_gpu) for samples in transposed]
+    elif isinstance(batch[0], Mapping):
+        return {
+            key: collate([d[key] for d in batch], samples_per_gpu)
+            for key in batch[0]
+        }
+    else:
+        return default_collate(batch)
diff --git a/annotator/uniformer/mmcv/parallel/data_container.py b/annotator/uniformer/mmcv/parallel/data_container.py
new file mode 100644
index 0000000000000000000000000000000000000000..cedb0d32a51a1f575a622b38de2cee3ab4757821
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/data_container.py
@@ -0,0 +1,89 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import functools
+
+import torch
+
+
+def assert_tensor_type(func):
+
+    @functools.wraps(func)
+    def wrapper(*args, **kwargs):
+        if not isinstance(args[0].data, torch.Tensor):
+            raise AttributeError(
+                f'{args[0].__class__.__name__} has no attribute '
+                f'{func.__name__} for type {args[0].datatype}')
+        return func(*args, **kwargs)
+
+    return wrapper
+
+
+class DataContainer:
+    """A container for any type of objects.
+
+    Typically tensors will be stacked in the collate function and sliced along
+    some dimension in the scatter function. This behavior has some limitations.
+    1. All tensors have to be the same size.
+    2. Types are limited (numpy array or Tensor).
+
+    We design `DataContainer` and `MMDataParallel` to overcome these
+    limitations. The behavior can be either of the following.
+
+    - copy to GPU, pad all tensors to the same size and stack them
+    - copy to GPU without stacking
+    - leave the objects as is and pass it to the model
+    - pad_dims specifies the number of last few dimensions to do padding
+    """
+
+    def __init__(self,
+                 data,
+                 stack=False,
+                 padding_value=0,
+                 cpu_only=False,
+                 pad_dims=2):
+        self._data = data
+        self._cpu_only = cpu_only
+        self._stack = stack
+        self._padding_value = padding_value
+        assert pad_dims in [None, 1, 2, 3]
+        self._pad_dims = pad_dims
+
+    def __repr__(self):
+        return f'{self.__class__.__name__}({repr(self.data)})'
+
+    def __len__(self):
+        return len(self._data)
+
+    @property
+    def data(self):
+        return self._data
+
+    @property
+    def datatype(self):
+        if isinstance(self.data, torch.Tensor):
+            return self.data.type()
+        else:
+            return type(self.data)
+
+    @property
+    def cpu_only(self):
+        return self._cpu_only
+
+    @property
+    def stack(self):
+        return self._stack
+
+    @property
+    def padding_value(self):
+        return self._padding_value
+
+    @property
+    def pad_dims(self):
+        return self._pad_dims
+
+    @assert_tensor_type
+    def size(self, *args, **kwargs):
+        return self.data.size(*args, **kwargs)
+
+    @assert_tensor_type
+    def dim(self):
+        return self.data.dim()
diff --git a/annotator/uniformer/mmcv/parallel/data_parallel.py b/annotator/uniformer/mmcv/parallel/data_parallel.py
new file mode 100644
index 0000000000000000000000000000000000000000..79b5f69b654cf647dc7ae9174223781ab5c607d2
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/data_parallel.py
@@ -0,0 +1,89 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from itertools import chain
+
+from torch.nn.parallel import DataParallel
+
+from .scatter_gather import scatter_kwargs
+
+
+class MMDataParallel(DataParallel):
+    """The DataParallel module that supports DataContainer.
+
+    MMDataParallel has two main differences with PyTorch DataParallel:
+
+    - It supports a custom type :class:`DataContainer` which allows more
+      flexible control of input data during both GPU and CPU inference.
+    - It implement two more APIs ``train_step()`` and ``val_step()``.
+
+    Args:
+        module (:class:`nn.Module`): Module to be encapsulated.
+        device_ids (list[int]): Device IDS of modules to be scattered to.
+            Defaults to None when GPU is not available.
+        output_device (str | int): Device ID for output. Defaults to None.
+        dim (int): Dimension used to scatter the data. Defaults to 0.
+    """
+
+    def __init__(self, *args, dim=0, **kwargs):
+        super(MMDataParallel, self).__init__(*args, dim=dim, **kwargs)
+        self.dim = dim
+
+    def forward(self, *inputs, **kwargs):
+        """Override the original forward function.
+
+        The main difference lies in the CPU inference where the data in
+        :class:`DataContainers` will still be gathered.
+        """
+        if not self.device_ids:
+            # We add the following line thus the module could gather and
+            # convert data containers as those in GPU inference
+            inputs, kwargs = self.scatter(inputs, kwargs, [-1])
+            return self.module(*inputs[0], **kwargs[0])
+        else:
+            return super().forward(*inputs, **kwargs)
+
+    def scatter(self, inputs, kwargs, device_ids):
+        return scatter_kwargs(inputs, kwargs, device_ids, dim=self.dim)
+
+    def train_step(self, *inputs, **kwargs):
+        if not self.device_ids:
+            # We add the following line thus the module could gather and
+            # convert data containers as those in GPU inference
+            inputs, kwargs = self.scatter(inputs, kwargs, [-1])
+            return self.module.train_step(*inputs[0], **kwargs[0])
+
+        assert len(self.device_ids) == 1, \
+            ('MMDataParallel only supports single GPU training, if you need to'
+             ' train with multiple GPUs, please use MMDistributedDataParallel'
+             'instead.')
+
+        for t in chain(self.module.parameters(), self.module.buffers()):
+            if t.device != self.src_device_obj:
+                raise RuntimeError(
+                    'module must have its parameters and buffers '
+                    f'on device {self.src_device_obj} (device_ids[0]) but '
+                    f'found one of them on device: {t.device}')
+
+        inputs, kwargs = self.scatter(inputs, kwargs, self.device_ids)
+        return self.module.train_step(*inputs[0], **kwargs[0])
+
+    def val_step(self, *inputs, **kwargs):
+        if not self.device_ids:
+            # We add the following line thus the module could gather and
+            # convert data containers as those in GPU inference
+            inputs, kwargs = self.scatter(inputs, kwargs, [-1])
+            return self.module.val_step(*inputs[0], **kwargs[0])
+
+        assert len(self.device_ids) == 1, \
+            ('MMDataParallel only supports single GPU training, if you need to'
+             ' train with multiple GPUs, please use MMDistributedDataParallel'
+             ' instead.')
+
+        for t in chain(self.module.parameters(), self.module.buffers()):
+            if t.device != self.src_device_obj:
+                raise RuntimeError(
+                    'module must have its parameters and buffers '
+                    f'on device {self.src_device_obj} (device_ids[0]) but '
+                    f'found one of them on device: {t.device}')
+
+        inputs, kwargs = self.scatter(inputs, kwargs, self.device_ids)
+        return self.module.val_step(*inputs[0], **kwargs[0])
diff --git a/annotator/uniformer/mmcv/parallel/distributed.py b/annotator/uniformer/mmcv/parallel/distributed.py
new file mode 100644
index 0000000000000000000000000000000000000000..1e4c27903db58a54d37ea1ed9ec0104098b486f2
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/distributed.py
@@ -0,0 +1,112 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch.nn.parallel.distributed import (DistributedDataParallel,
+                                           _find_tensors)
+
+from annotator.uniformer.mmcv import print_log
+from annotator.uniformer.mmcv.utils import TORCH_VERSION, digit_version
+from .scatter_gather import scatter_kwargs
+
+
+class MMDistributedDataParallel(DistributedDataParallel):
+    """The DDP module that supports DataContainer.
+
+    MMDDP has two main differences with PyTorch DDP:
+
+    - It supports a custom type :class:`DataContainer` which allows more
+      flexible control of input data.
+    - It implement two APIs ``train_step()`` and ``val_step()``.
+    """
+
+    def to_kwargs(self, inputs, kwargs, device_id):
+        # Use `self.to_kwargs` instead of `self.scatter` in pytorch1.8
+        # to move all tensors to device_id
+        return scatter_kwargs(inputs, kwargs, [device_id], dim=self.dim)
+
+    def scatter(self, inputs, kwargs, device_ids):
+        return scatter_kwargs(inputs, kwargs, device_ids, dim=self.dim)
+
+    def train_step(self, *inputs, **kwargs):
+        """train_step() API for module wrapped by DistributedDataParallel.
+
+        This method is basically the same as
+        ``DistributedDataParallel.forward()``, while replacing
+        ``self.module.forward()`` with ``self.module.train_step()``.
+        It is compatible with PyTorch 1.1 - 1.5.
+        """
+
+        # In PyTorch >= 1.7, ``reducer._rebuild_buckets()`` is moved from the
+        # end of backward to the beginning of forward.
+        if ('parrots' not in TORCH_VERSION
+                and digit_version(TORCH_VERSION) >= digit_version('1.7')
+                and self.reducer._rebuild_buckets()):
+            print_log(
+                'Reducer buckets have been rebuilt in this iteration.',
+                logger='mmcv')
+
+        if getattr(self, 'require_forward_param_sync', True):
+            self._sync_params()
+        if self.device_ids:
+            inputs, kwargs = self.scatter(inputs, kwargs, self.device_ids)
+            if len(self.device_ids) == 1:
+                output = self.module.train_step(*inputs[0], **kwargs[0])
+            else:
+                outputs = self.parallel_apply(
+                    self._module_copies[:len(inputs)], inputs, kwargs)
+                output = self.gather(outputs, self.output_device)
+        else:
+            output = self.module.train_step(*inputs, **kwargs)
+
+        if torch.is_grad_enabled() and getattr(
+                self, 'require_backward_grad_sync', True):
+            if self.find_unused_parameters:
+                self.reducer.prepare_for_backward(list(_find_tensors(output)))
+            else:
+                self.reducer.prepare_for_backward([])
+        else:
+            if ('parrots' not in TORCH_VERSION
+                    and digit_version(TORCH_VERSION) > digit_version('1.2')):
+                self.require_forward_param_sync = False
+        return output
+
+    def val_step(self, *inputs, **kwargs):
+        """val_step() API for module wrapped by DistributedDataParallel.
+
+        This method is basically the same as
+        ``DistributedDataParallel.forward()``, while replacing
+        ``self.module.forward()`` with ``self.module.val_step()``.
+        It is compatible with PyTorch 1.1 - 1.5.
+        """
+        # In PyTorch >= 1.7, ``reducer._rebuild_buckets()`` is moved from the
+        # end of backward to the beginning of forward.
+        if ('parrots' not in TORCH_VERSION
+                and digit_version(TORCH_VERSION) >= digit_version('1.7')
+                and self.reducer._rebuild_buckets()):
+            print_log(
+                'Reducer buckets have been rebuilt in this iteration.',
+                logger='mmcv')
+
+        if getattr(self, 'require_forward_param_sync', True):
+            self._sync_params()
+        if self.device_ids:
+            inputs, kwargs = self.scatter(inputs, kwargs, self.device_ids)
+            if len(self.device_ids) == 1:
+                output = self.module.val_step(*inputs[0], **kwargs[0])
+            else:
+                outputs = self.parallel_apply(
+                    self._module_copies[:len(inputs)], inputs, kwargs)
+                output = self.gather(outputs, self.output_device)
+        else:
+            output = self.module.val_step(*inputs, **kwargs)
+
+        if torch.is_grad_enabled() and getattr(
+                self, 'require_backward_grad_sync', True):
+            if self.find_unused_parameters:
+                self.reducer.prepare_for_backward(list(_find_tensors(output)))
+            else:
+                self.reducer.prepare_for_backward([])
+        else:
+            if ('parrots' not in TORCH_VERSION
+                    and digit_version(TORCH_VERSION) > digit_version('1.2')):
+                self.require_forward_param_sync = False
+        return output
diff --git a/annotator/uniformer/mmcv/parallel/distributed_deprecated.py b/annotator/uniformer/mmcv/parallel/distributed_deprecated.py
new file mode 100644
index 0000000000000000000000000000000000000000..676937a2085d4da20fa87923041a200fca6214eb
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/distributed_deprecated.py
@@ -0,0 +1,70 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+import torch.distributed as dist
+import torch.nn as nn
+from torch._utils import (_flatten_dense_tensors, _take_tensors,
+                          _unflatten_dense_tensors)
+
+from annotator.uniformer.mmcv.utils import TORCH_VERSION, digit_version
+from .registry import MODULE_WRAPPERS
+from .scatter_gather import scatter_kwargs
+
+
+@MODULE_WRAPPERS.register_module()
+class MMDistributedDataParallel(nn.Module):
+
+    def __init__(self,
+                 module,
+                 dim=0,
+                 broadcast_buffers=True,
+                 bucket_cap_mb=25):
+        super(MMDistributedDataParallel, self).__init__()
+        self.module = module
+        self.dim = dim
+        self.broadcast_buffers = broadcast_buffers
+
+        self.broadcast_bucket_size = bucket_cap_mb * 1024 * 1024
+        self._sync_params()
+
+    def _dist_broadcast_coalesced(self, tensors, buffer_size):
+        for tensors in _take_tensors(tensors, buffer_size):
+            flat_tensors = _flatten_dense_tensors(tensors)
+            dist.broadcast(flat_tensors, 0)
+            for tensor, synced in zip(
+                    tensors, _unflatten_dense_tensors(flat_tensors, tensors)):
+                tensor.copy_(synced)
+
+    def _sync_params(self):
+        module_states = list(self.module.state_dict().values())
+        if len(module_states) > 0:
+            self._dist_broadcast_coalesced(module_states,
+                                           self.broadcast_bucket_size)
+        if self.broadcast_buffers:
+            if (TORCH_VERSION != 'parrots'
+                    and digit_version(TORCH_VERSION) < digit_version('1.0')):
+                buffers = [b.data for b in self.module._all_buffers()]
+            else:
+                buffers = [b.data for b in self.module.buffers()]
+            if len(buffers) > 0:
+                self._dist_broadcast_coalesced(buffers,
+                                               self.broadcast_bucket_size)
+
+    def scatter(self, inputs, kwargs, device_ids):
+        return scatter_kwargs(inputs, kwargs, device_ids, dim=self.dim)
+
+    def forward(self, *inputs, **kwargs):
+        inputs, kwargs = self.scatter(inputs, kwargs,
+                                      [torch.cuda.current_device()])
+        return self.module(*inputs[0], **kwargs[0])
+
+    def train_step(self, *inputs, **kwargs):
+        inputs, kwargs = self.scatter(inputs, kwargs,
+                                      [torch.cuda.current_device()])
+        output = self.module.train_step(*inputs[0], **kwargs[0])
+        return output
+
+    def val_step(self, *inputs, **kwargs):
+        inputs, kwargs = self.scatter(inputs, kwargs,
+                                      [torch.cuda.current_device()])
+        output = self.module.val_step(*inputs[0], **kwargs[0])
+        return output
diff --git a/annotator/uniformer/mmcv/parallel/registry.py b/annotator/uniformer/mmcv/parallel/registry.py
new file mode 100644
index 0000000000000000000000000000000000000000..a204a07fba10e614223f090d1a57cf9c4d74d4a1
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/registry.py
@@ -0,0 +1,8 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from torch.nn.parallel import DataParallel, DistributedDataParallel
+
+from annotator.uniformer.mmcv.utils import Registry
+
+MODULE_WRAPPERS = Registry('module wrapper')
+MODULE_WRAPPERS.register_module(module=DataParallel)
+MODULE_WRAPPERS.register_module(module=DistributedDataParallel)
diff --git a/annotator/uniformer/mmcv/parallel/scatter_gather.py b/annotator/uniformer/mmcv/parallel/scatter_gather.py
new file mode 100644
index 0000000000000000000000000000000000000000..900ff88566f8f14830590459dc4fd16d4b382e47
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/scatter_gather.py
@@ -0,0 +1,59 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from torch.nn.parallel._functions import Scatter as OrigScatter
+
+from ._functions import Scatter
+from .data_container import DataContainer
+
+
+def scatter(inputs, target_gpus, dim=0):
+    """Scatter inputs to target gpus.
+
+    The only difference from original :func:`scatter` is to add support for
+    :type:`~mmcv.parallel.DataContainer`.
+    """
+
+    def scatter_map(obj):
+        if isinstance(obj, torch.Tensor):
+            if target_gpus != [-1]:
+                return OrigScatter.apply(target_gpus, None, dim, obj)
+            else:
+                # for CPU inference we use self-implemented scatter
+                return Scatter.forward(target_gpus, obj)
+        if isinstance(obj, DataContainer):
+            if obj.cpu_only:
+                return obj.data
+            else:
+                return Scatter.forward(target_gpus, obj.data)
+        if isinstance(obj, tuple) and len(obj) > 0:
+            return list(zip(*map(scatter_map, obj)))
+        if isinstance(obj, list) and len(obj) > 0:
+            out = list(map(list, zip(*map(scatter_map, obj))))
+            return out
+        if isinstance(obj, dict) and len(obj) > 0:
+            out = list(map(type(obj), zip(*map(scatter_map, obj.items()))))
+            return out
+        return [obj for targets in target_gpus]
+
+    # After scatter_map is called, a scatter_map cell will exist. This cell
+    # has a reference to the actual function scatter_map, which has references
+    # to a closure that has a reference to the scatter_map cell (because the
+    # fn is recursive). To avoid this reference cycle, we set the function to
+    # None, clearing the cell
+    try:
+        return scatter_map(inputs)
+    finally:
+        scatter_map = None
+
+
+def scatter_kwargs(inputs, kwargs, target_gpus, dim=0):
+    """Scatter with support for kwargs dictionary."""
+    inputs = scatter(inputs, target_gpus, dim) if inputs else []
+    kwargs = scatter(kwargs, target_gpus, dim) if kwargs else []
+    if len(inputs) < len(kwargs):
+        inputs.extend([() for _ in range(len(kwargs) - len(inputs))])
+    elif len(kwargs) < len(inputs):
+        kwargs.extend([{} for _ in range(len(inputs) - len(kwargs))])
+    inputs = tuple(inputs)
+    kwargs = tuple(kwargs)
+    return inputs, kwargs
diff --git a/annotator/uniformer/mmcv/parallel/utils.py b/annotator/uniformer/mmcv/parallel/utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..0f5712cb42c38a2e8563bf563efb6681383cab9b
--- /dev/null
+++ b/annotator/uniformer/mmcv/parallel/utils.py
@@ -0,0 +1,20 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .registry import MODULE_WRAPPERS
+
+
+def is_module_wrapper(module):
+    """Check if a module is a module wrapper.
+
+    The following 3 modules in MMCV (and their subclasses) are regarded as
+    module wrappers: DataParallel, DistributedDataParallel,
+    MMDistributedDataParallel (the deprecated version). You may add you own
+    module wrapper by registering it to mmcv.parallel.MODULE_WRAPPERS.
+
+    Args:
+        module (nn.Module): The module to be checked.
+
+    Returns:
+        bool: True if the input module is a module wrapper.
+    """
+    module_wrappers = tuple(MODULE_WRAPPERS.module_dict.values())
+    return isinstance(module, module_wrappers)
diff --git a/annotator/uniformer/mmcv/runner/__init__.py b/annotator/uniformer/mmcv/runner/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..52e4b48d383a84a055dcd7f6236f6e8e58eab924
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/__init__.py
@@ -0,0 +1,47 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .base_module import BaseModule, ModuleList, Sequential
+from .base_runner import BaseRunner
+from .builder import RUNNERS, build_runner
+from .checkpoint import (CheckpointLoader, _load_checkpoint,
+                         _load_checkpoint_with_prefix, load_checkpoint,
+                         load_state_dict, save_checkpoint, weights_to_cpu)
+from .default_constructor import DefaultRunnerConstructor
+from .dist_utils import (allreduce_grads, allreduce_params, get_dist_info,
+                         init_dist, master_only)
+from .epoch_based_runner import EpochBasedRunner, Runner
+from .fp16_utils import LossScaler, auto_fp16, force_fp32, wrap_fp16_model
+from .hooks import (HOOKS, CheckpointHook, ClosureHook, DistEvalHook,
+                    DistSamplerSeedHook, DvcliveLoggerHook, EMAHook, EvalHook,
+                    Fp16OptimizerHook, GradientCumulativeFp16OptimizerHook,
+                    GradientCumulativeOptimizerHook, Hook, IterTimerHook,
+                    LoggerHook, LrUpdaterHook, MlflowLoggerHook,
+                    NeptuneLoggerHook, OptimizerHook, PaviLoggerHook,
+                    SyncBuffersHook, TensorboardLoggerHook, TextLoggerHook,
+                    WandbLoggerHook)
+from .iter_based_runner import IterBasedRunner, IterLoader
+from .log_buffer import LogBuffer
+from .optimizer import (OPTIMIZER_BUILDERS, OPTIMIZERS,
+                        DefaultOptimizerConstructor, build_optimizer,
+                        build_optimizer_constructor)
+from .priority import Priority, get_priority
+from .utils import get_host_info, get_time_str, obj_from_dict, set_random_seed
+
+__all__ = [
+    'BaseRunner', 'Runner', 'EpochBasedRunner', 'IterBasedRunner', 'LogBuffer',
+    'HOOKS', 'Hook', 'CheckpointHook', 'ClosureHook', 'LrUpdaterHook',
+    'OptimizerHook', 'IterTimerHook', 'DistSamplerSeedHook', 'LoggerHook',
+    'PaviLoggerHook', 'TextLoggerHook', 'TensorboardLoggerHook',
+    'NeptuneLoggerHook', 'WandbLoggerHook', 'MlflowLoggerHook',
+    'DvcliveLoggerHook', '_load_checkpoint', 'load_state_dict',
+    'load_checkpoint', 'weights_to_cpu', 'save_checkpoint', 'Priority',
+    'get_priority', 'get_host_info', 'get_time_str', 'obj_from_dict',
+    'init_dist', 'get_dist_info', 'master_only', 'OPTIMIZER_BUILDERS',
+    'OPTIMIZERS', 'DefaultOptimizerConstructor', 'build_optimizer',
+    'build_optimizer_constructor', 'IterLoader', 'set_random_seed',
+    'auto_fp16', 'force_fp32', 'wrap_fp16_model', 'Fp16OptimizerHook',
+    'SyncBuffersHook', 'EMAHook', 'build_runner', 'RUNNERS', 'allreduce_grads',
+    'allreduce_params', 'LossScaler', 'CheckpointLoader', 'BaseModule',
+    '_load_checkpoint_with_prefix', 'EvalHook', 'DistEvalHook', 'Sequential',
+    'ModuleList', 'GradientCumulativeOptimizerHook',
+    'GradientCumulativeFp16OptimizerHook', 'DefaultRunnerConstructor'
+]
diff --git a/annotator/uniformer/mmcv/runner/base_module.py b/annotator/uniformer/mmcv/runner/base_module.py
new file mode 100644
index 0000000000000000000000000000000000000000..617fad9bb89f10a9a0911d962dfb3bc8f3a3628c
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/base_module.py
@@ -0,0 +1,195 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import copy
+import warnings
+from abc import ABCMeta
+from collections import defaultdict
+from logging import FileHandler
+
+import torch.nn as nn
+
+from annotator.uniformer.mmcv.runner.dist_utils import master_only
+from annotator.uniformer.mmcv.utils.logging import get_logger, logger_initialized, print_log
+
+
+class BaseModule(nn.Module, metaclass=ABCMeta):
+    """Base module for all modules in openmmlab.
+
+    ``BaseModule`` is a wrapper of ``torch.nn.Module`` with additional
+    functionality of parameter initialization. Compared with
+    ``torch.nn.Module``, ``BaseModule`` mainly adds three attributes.
+
+        - ``init_cfg``: the config to control the initialization.
+        - ``init_weights``: The function of parameter
+            initialization and recording initialization
+            information.
+        - ``_params_init_info``: Used to track the parameter
+            initialization information. This attribute only
+            exists during executing the ``init_weights``.
+
+    Args:
+        init_cfg (dict, optional): Initialization config dict.
+    """
+
+    def __init__(self, init_cfg=None):
+        """Initialize BaseModule, inherited from `torch.nn.Module`"""
+
+        # NOTE init_cfg can be defined in different levels, but init_cfg
+        # in low levels has a higher priority.
+
+        super(BaseModule, self).__init__()
+        # define default value of init_cfg instead of hard code
+        # in init_weights() function
+        self._is_init = False
+
+        self.init_cfg = copy.deepcopy(init_cfg)
+
+        # Backward compatibility in derived classes
+        # if pretrained is not None:
+        #     warnings.warn('DeprecationWarning: pretrained is a deprecated \
+        #         key, please consider using init_cfg')
+        #     self.init_cfg = dict(type='Pretrained', checkpoint=pretrained)
+
+    @property
+    def is_init(self):
+        return self._is_init
+
+    def init_weights(self):
+        """Initialize the weights."""
+
+        is_top_level_module = False
+        # check if it is top-level module
+        if not hasattr(self, '_params_init_info'):
+            # The `_params_init_info` is used to record the initialization
+            # information of the parameters
+            # the key should be the obj:`nn.Parameter` of model and the value
+            # should be a dict containing
+            # - init_info (str): The string that describes the initialization.
+            # - tmp_mean_value (FloatTensor): The mean of the parameter,
+            #       which indicates whether the parameter has been modified.
+            # this attribute would be deleted after all parameters
+            # is initialized.
+            self._params_init_info = defaultdict(dict)
+            is_top_level_module = True
+
+            # Initialize the `_params_init_info`,
+            # When detecting the `tmp_mean_value` of
+            # the corresponding parameter is changed, update related
+            # initialization information
+            for name, param in self.named_parameters():
+                self._params_init_info[param][
+                    'init_info'] = f'The value is the same before and ' \
+                                   f'after calling `init_weights` ' \
+                                   f'of {self.__class__.__name__} '
+                self._params_init_info[param][
+                    'tmp_mean_value'] = param.data.mean()
+
+            # pass `params_init_info` to all submodules
+            # All submodules share the same `params_init_info`,
+            # so it will be updated when parameters are
+            # modified at any level of the model.
+            for sub_module in self.modules():
+                sub_module._params_init_info = self._params_init_info
+
+        # Get the initialized logger, if not exist,
+        # create a logger named `mmcv`
+        logger_names = list(logger_initialized.keys())
+        logger_name = logger_names[0] if logger_names else 'mmcv'
+
+        from ..cnn import initialize
+        from ..cnn.utils.weight_init import update_init_info
+        module_name = self.__class__.__name__
+        if not self._is_init:
+            if self.init_cfg:
+                print_log(
+                    f'initialize {module_name} with init_cfg {self.init_cfg}',
+                    logger=logger_name)
+                initialize(self, self.init_cfg)
+                if isinstance(self.init_cfg, dict):
+                    # prevent the parameters of
+                    # the pre-trained model
+                    # from being overwritten by
+                    # the `init_weights`
+                    if self.init_cfg['type'] == 'Pretrained':
+                        return
+
+            for m in self.children():
+                if hasattr(m, 'init_weights'):
+                    m.init_weights()
+                    # users may overload the `init_weights`
+                    update_init_info(
+                        m,
+                        init_info=f'Initialized by '
+                        f'user-defined `init_weights`'
+                        f' in {m.__class__.__name__} ')
+
+            self._is_init = True
+        else:
+            warnings.warn(f'init_weights of {self.__class__.__name__} has '
+                          f'been called more than once.')
+
+        if is_top_level_module:
+            self._dump_init_info(logger_name)
+
+            for sub_module in self.modules():
+                del sub_module._params_init_info
+
+    @master_only
+    def _dump_init_info(self, logger_name):
+        """Dump the initialization information to a file named
+        `initialization.log.json` in workdir.
+
+        Args:
+            logger_name (str): The name of logger.
+        """
+
+        logger = get_logger(logger_name)
+
+        with_file_handler = False
+        # dump the information to the logger file if there is a `FileHandler`
+        for handler in logger.handlers:
+            if isinstance(handler, FileHandler):
+                handler.stream.write(
+                    'Name of parameter - Initialization information\n')
+                for name, param in self.named_parameters():
+                    handler.stream.write(
+                        f'\n{name} - {param.shape}: '
+                        f"\n{self._params_init_info[param]['init_info']} \n")
+                handler.stream.flush()
+                with_file_handler = True
+        if not with_file_handler:
+            for name, param in self.named_parameters():
+                print_log(
+                    f'\n{name} - {param.shape}: '
+                    f"\n{self._params_init_info[param]['init_info']} \n ",
+                    logger=logger_name)
+
+    def __repr__(self):
+        s = super().__repr__()
+        if self.init_cfg:
+            s += f'\ninit_cfg={self.init_cfg}'
+        return s
+
+
+class Sequential(BaseModule, nn.Sequential):
+    """Sequential module in openmmlab.
+
+    Args:
+        init_cfg (dict, optional): Initialization config dict.
+    """
+
+    def __init__(self, *args, init_cfg=None):
+        BaseModule.__init__(self, init_cfg)
+        nn.Sequential.__init__(self, *args)
+
+
+class ModuleList(BaseModule, nn.ModuleList):
+    """ModuleList in openmmlab.
+
+    Args:
+        modules (iterable, optional): an iterable of modules to add.
+        init_cfg (dict, optional): Initialization config dict.
+    """
+
+    def __init__(self, modules=None, init_cfg=None):
+        BaseModule.__init__(self, init_cfg)
+        nn.ModuleList.__init__(self, modules)
diff --git a/annotator/uniformer/mmcv/runner/base_runner.py b/annotator/uniformer/mmcv/runner/base_runner.py
new file mode 100644
index 0000000000000000000000000000000000000000..4928db0a73b56fe0218a4bf66ec4ffa082d31ccc
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/base_runner.py
@@ -0,0 +1,542 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import copy
+import logging
+import os.path as osp
+import warnings
+from abc import ABCMeta, abstractmethod
+
+import torch
+from torch.optim import Optimizer
+
+import annotator.uniformer.mmcv as mmcv
+from ..parallel import is_module_wrapper
+from .checkpoint import load_checkpoint
+from .dist_utils import get_dist_info
+from .hooks import HOOKS, Hook
+from .log_buffer import LogBuffer
+from .priority import Priority, get_priority
+from .utils import get_time_str
+
+
+class BaseRunner(metaclass=ABCMeta):
+    """The base class of Runner, a training helper for PyTorch.
+
+    All subclasses should implement the following APIs:
+
+    - ``run()``
+    - ``train()``
+    - ``val()``
+    - ``save_checkpoint()``
+
+    Args:
+        model (:obj:`torch.nn.Module`): The model to be run.
+        batch_processor (callable): A callable method that process a data
+            batch. The interface of this method should be
+            `batch_processor(model, data, train_mode) -> dict`
+        optimizer (dict or :obj:`torch.optim.Optimizer`): It can be either an
+            optimizer (in most cases) or a dict of optimizers (in models that
+            requires more than one optimizer, e.g., GAN).
+        work_dir (str, optional): The working directory to save checkpoints
+            and logs. Defaults to None.
+        logger (:obj:`logging.Logger`): Logger used during training.
+             Defaults to None. (The default value is just for backward
+             compatibility)
+        meta (dict | None): A dict records some import information such as
+            environment info and seed, which will be logged in logger hook.
+            Defaults to None.
+        max_epochs (int, optional): Total training epochs.
+        max_iters (int, optional): Total training iterations.
+    """
+
+    def __init__(self,
+                 model,
+                 batch_processor=None,
+                 optimizer=None,
+                 work_dir=None,
+                 logger=None,
+                 meta=None,
+                 max_iters=None,
+                 max_epochs=None):
+        if batch_processor is not None:
+            if not callable(batch_processor):
+                raise TypeError('batch_processor must be callable, '
+                                f'but got {type(batch_processor)}')
+            warnings.warn('batch_processor is deprecated, please implement '
+                          'train_step() and val_step() in the model instead.')
+            # raise an error is `batch_processor` is not None and
+            # `model.train_step()` exists.
+            if is_module_wrapper(model):
+                _model = model.module
+            else:
+                _model = model
+            if hasattr(_model, 'train_step') or hasattr(_model, 'val_step'):
+                raise RuntimeError(
+                    'batch_processor and model.train_step()/model.val_step() '
+                    'cannot be both available.')
+        else:
+            assert hasattr(model, 'train_step')
+
+        # check the type of `optimizer`
+        if isinstance(optimizer, dict):
+            for name, optim in optimizer.items():
+                if not isinstance(optim, Optimizer):
+                    raise TypeError(
+                        f'optimizer must be a dict of torch.optim.Optimizers, '
+                        f'but optimizer["{name}"] is a {type(optim)}')
+        elif not isinstance(optimizer, Optimizer) and optimizer is not None:
+            raise TypeError(
+                f'optimizer must be a torch.optim.Optimizer object '
+                f'or dict or None, but got {type(optimizer)}')
+
+        # check the type of `logger`
+        if not isinstance(logger, logging.Logger):
+            raise TypeError(f'logger must be a logging.Logger object, '
+                            f'but got {type(logger)}')
+
+        # check the type of `meta`
+        if meta is not None and not isinstance(meta, dict):
+            raise TypeError(
+                f'meta must be a dict or None, but got {type(meta)}')
+
+        self.model = model
+        self.batch_processor = batch_processor
+        self.optimizer = optimizer
+        self.logger = logger
+        self.meta = meta
+        # create work_dir
+        if mmcv.is_str(work_dir):
+            self.work_dir = osp.abspath(work_dir)
+            mmcv.mkdir_or_exist(self.work_dir)
+        elif work_dir is None:
+            self.work_dir = None
+        else:
+            raise TypeError('"work_dir" must be a str or None')
+
+        # get model name from the model class
+        if hasattr(self.model, 'module'):
+            self._model_name = self.model.module.__class__.__name__
+        else:
+            self._model_name = self.model.__class__.__name__
+
+        self._rank, self._world_size = get_dist_info()
+        self.timestamp = get_time_str()
+        self.mode = None
+        self._hooks = []
+        self._epoch = 0
+        self._iter = 0
+        self._inner_iter = 0
+
+        if max_epochs is not None and max_iters is not None:
+            raise ValueError(
+                'Only one of `max_epochs` or `max_iters` can be set.')
+
+        self._max_epochs = max_epochs
+        self._max_iters = max_iters
+        # TODO: Redesign LogBuffer, it is not flexible and elegant enough
+        self.log_buffer = LogBuffer()
+
+    @property
+    def model_name(self):
+        """str: Name of the model, usually the module class name."""
+        return self._model_name
+
+    @property
+    def rank(self):
+        """int: Rank of current process. (distributed training)"""
+        return self._rank
+
+    @property
+    def world_size(self):
+        """int: Number of processes participating in the job.
+        (distributed training)"""
+        return self._world_size
+
+    @property
+    def hooks(self):
+        """list[:obj:`Hook`]: A list of registered hooks."""
+        return self._hooks
+
+    @property
+    def epoch(self):
+        """int: Current epoch."""
+        return self._epoch
+
+    @property
+    def iter(self):
+        """int: Current iteration."""
+        return self._iter
+
+    @property
+    def inner_iter(self):
+        """int: Iteration in an epoch."""
+        return self._inner_iter
+
+    @property
+    def max_epochs(self):
+        """int: Maximum training epochs."""
+        return self._max_epochs
+
+    @property
+    def max_iters(self):
+        """int: Maximum training iterations."""
+        return self._max_iters
+
+    @abstractmethod
+    def train(self):
+        pass
+
+    @abstractmethod
+    def val(self):
+        pass
+
+    @abstractmethod
+    def run(self, data_loaders, workflow, **kwargs):
+        pass
+
+    @abstractmethod
+    def save_checkpoint(self,
+                        out_dir,
+                        filename_tmpl,
+                        save_optimizer=True,
+                        meta=None,
+                        create_symlink=True):
+        pass
+
+    def current_lr(self):
+        """Get current learning rates.
+
+        Returns:
+            list[float] | dict[str, list[float]]: Current learning rates of all
+                param groups. If the runner has a dict of optimizers, this
+                method will return a dict.
+        """
+        if isinstance(self.optimizer, torch.optim.Optimizer):
+            lr = [group['lr'] for group in self.optimizer.param_groups]
+        elif isinstance(self.optimizer, dict):
+            lr = dict()
+            for name, optim in self.optimizer.items():
+                lr[name] = [group['lr'] for group in optim.param_groups]
+        else:
+            raise RuntimeError(
+                'lr is not applicable because optimizer does not exist.')
+        return lr
+
+    def current_momentum(self):
+        """Get current momentums.
+
+        Returns:
+            list[float] | dict[str, list[float]]: Current momentums of all
+                param groups. If the runner has a dict of optimizers, this
+                method will return a dict.
+        """
+
+        def _get_momentum(optimizer):
+            momentums = []
+            for group in optimizer.param_groups:
+                if 'momentum' in group.keys():
+                    momentums.append(group['momentum'])
+                elif 'betas' in group.keys():
+                    momentums.append(group['betas'][0])
+                else:
+                    momentums.append(0)
+            return momentums
+
+        if self.optimizer is None:
+            raise RuntimeError(
+                'momentum is not applicable because optimizer does not exist.')
+        elif isinstance(self.optimizer, torch.optim.Optimizer):
+            momentums = _get_momentum(self.optimizer)
+        elif isinstance(self.optimizer, dict):
+            momentums = dict()
+            for name, optim in self.optimizer.items():
+                momentums[name] = _get_momentum(optim)
+        return momentums
+
+    def register_hook(self, hook, priority='NORMAL'):
+        """Register a hook into the hook list.
+
+        The hook will be inserted into a priority queue, with the specified
+        priority (See :class:`Priority` for details of priorities).
+        For hooks with the same priority, they will be triggered in the same
+        order as they are registered.
+
+        Args:
+            hook (:obj:`Hook`): The hook to be registered.
+            priority (int or str or :obj:`Priority`): Hook priority.
+                Lower value means higher priority.
+        """
+        assert isinstance(hook, Hook)
+        if hasattr(hook, 'priority'):
+            raise ValueError('"priority" is a reserved attribute for hooks')
+        priority = get_priority(priority)
+        hook.priority = priority
+        # insert the hook to a sorted list
+        inserted = False
+        for i in range(len(self._hooks) - 1, -1, -1):
+            if priority >= self._hooks[i].priority:
+                self._hooks.insert(i + 1, hook)
+                inserted = True
+                break
+        if not inserted:
+            self._hooks.insert(0, hook)
+
+    def register_hook_from_cfg(self, hook_cfg):
+        """Register a hook from its cfg.
+
+        Args:
+            hook_cfg (dict): Hook config. It should have at least keys 'type'
+              and 'priority' indicating its type and priority.
+
+        Notes:
+            The specific hook class to register should not use 'type' and
+            'priority' arguments during initialization.
+        """
+        hook_cfg = hook_cfg.copy()
+        priority = hook_cfg.pop('priority', 'NORMAL')
+        hook = mmcv.build_from_cfg(hook_cfg, HOOKS)
+        self.register_hook(hook, priority=priority)
+
+    def call_hook(self, fn_name):
+        """Call all hooks.
+
+        Args:
+            fn_name (str): The function name in each hook to be called, such as
+                "before_train_epoch".
+        """
+        for hook in self._hooks:
+            getattr(hook, fn_name)(self)
+
+    def get_hook_info(self):
+        # Get hooks info in each stage
+        stage_hook_map = {stage: [] for stage in Hook.stages}
+        for hook in self.hooks:
+            try:
+                priority = Priority(hook.priority).name
+            except ValueError:
+                priority = hook.priority
+            classname = hook.__class__.__name__
+            hook_info = f'({priority:<12}) {classname:<35}'
+            for trigger_stage in hook.get_triggered_stages():
+                stage_hook_map[trigger_stage].append(hook_info)
+
+        stage_hook_infos = []
+        for stage in Hook.stages:
+            hook_infos = stage_hook_map[stage]
+            if len(hook_infos) > 0:
+                info = f'{stage}:\n'
+                info += '\n'.join(hook_infos)
+                info += '\n -------------------- '
+                stage_hook_infos.append(info)
+        return '\n'.join(stage_hook_infos)
+
+    def load_checkpoint(self,
+                        filename,
+                        map_location='cpu',
+                        strict=False,
+                        revise_keys=[(r'^module.', '')]):
+        return load_checkpoint(
+            self.model,
+            filename,
+            map_location,
+            strict,
+            self.logger,
+            revise_keys=revise_keys)
+
+    def resume(self,
+               checkpoint,
+               resume_optimizer=True,
+               map_location='default'):
+        if map_location == 'default':
+            if torch.cuda.is_available():
+                device_id = torch.cuda.current_device()
+                checkpoint = self.load_checkpoint(
+                    checkpoint,
+                    map_location=lambda storage, loc: storage.cuda(device_id))
+            else:
+                checkpoint = self.load_checkpoint(checkpoint)
+        else:
+            checkpoint = self.load_checkpoint(
+                checkpoint, map_location=map_location)
+
+        self._epoch = checkpoint['meta']['epoch']
+        self._iter = checkpoint['meta']['iter']
+        if self.meta is None:
+            self.meta = {}
+        self.meta.setdefault('hook_msgs', {})
+        # load `last_ckpt`, `best_score`, `best_ckpt`, etc. for hook messages
+        self.meta['hook_msgs'].update(checkpoint['meta'].get('hook_msgs', {}))
+
+        # Re-calculate the number of iterations when resuming
+        # models with different number of GPUs
+        if 'config' in checkpoint['meta']:
+            config = mmcv.Config.fromstring(
+                checkpoint['meta']['config'], file_format='.py')
+            previous_gpu_ids = config.get('gpu_ids', None)
+            if previous_gpu_ids and len(previous_gpu_ids) > 0 and len(
+                    previous_gpu_ids) != self.world_size:
+                self._iter = int(self._iter * len(previous_gpu_ids) /
+                                 self.world_size)
+                self.logger.info('the iteration number is changed due to '
+                                 'change of GPU number')
+
+        # resume meta information meta
+        self.meta = checkpoint['meta']
+
+        if 'optimizer' in checkpoint and resume_optimizer:
+            if isinstance(self.optimizer, Optimizer):
+                self.optimizer.load_state_dict(checkpoint['optimizer'])
+            elif isinstance(self.optimizer, dict):
+                for k in self.optimizer.keys():
+                    self.optimizer[k].load_state_dict(
+                        checkpoint['optimizer'][k])
+            else:
+                raise TypeError(
+                    'Optimizer should be dict or torch.optim.Optimizer '
+                    f'but got {type(self.optimizer)}')
+
+        self.logger.info('resumed epoch %d, iter %d', self.epoch, self.iter)
+
+    def register_lr_hook(self, lr_config):
+        if lr_config is None:
+            return
+        elif isinstance(lr_config, dict):
+            assert 'policy' in lr_config
+            policy_type = lr_config.pop('policy')
+            # If the type of policy is all in lower case, e.g., 'cyclic',
+            # then its first letter will be capitalized, e.g., to be 'Cyclic'.
+            # This is for the convenient usage of Lr updater.
+            # Since this is not applicable for `
+            # CosineAnnealingLrUpdater`,
+            # the string will not be changed if it contains capital letters.
+            if policy_type == policy_type.lower():
+                policy_type = policy_type.title()
+            hook_type = policy_type + 'LrUpdaterHook'
+            lr_config['type'] = hook_type
+            hook = mmcv.build_from_cfg(lr_config, HOOKS)
+        else:
+            hook = lr_config
+        self.register_hook(hook, priority='VERY_HIGH')
+
+    def register_momentum_hook(self, momentum_config):
+        if momentum_config is None:
+            return
+        if isinstance(momentum_config, dict):
+            assert 'policy' in momentum_config
+            policy_type = momentum_config.pop('policy')
+            # If the type of policy is all in lower case, e.g., 'cyclic',
+            # then its first letter will be capitalized, e.g., to be 'Cyclic'.
+            # This is for the convenient usage of momentum updater.
+            # Since this is not applicable for
+            # `CosineAnnealingMomentumUpdater`,
+            # the string will not be changed if it contains capital letters.
+            if policy_type == policy_type.lower():
+                policy_type = policy_type.title()
+            hook_type = policy_type + 'MomentumUpdaterHook'
+            momentum_config['type'] = hook_type
+            hook = mmcv.build_from_cfg(momentum_config, HOOKS)
+        else:
+            hook = momentum_config
+        self.register_hook(hook, priority='HIGH')
+
+    def register_optimizer_hook(self, optimizer_config):
+        if optimizer_config is None:
+            return
+        if isinstance(optimizer_config, dict):
+            optimizer_config.setdefault('type', 'OptimizerHook')
+            hook = mmcv.build_from_cfg(optimizer_config, HOOKS)
+        else:
+            hook = optimizer_config
+        self.register_hook(hook, priority='ABOVE_NORMAL')
+
+    def register_checkpoint_hook(self, checkpoint_config):
+        if checkpoint_config is None:
+            return
+        if isinstance(checkpoint_config, dict):
+            checkpoint_config.setdefault('type', 'CheckpointHook')
+            hook = mmcv.build_from_cfg(checkpoint_config, HOOKS)
+        else:
+            hook = checkpoint_config
+        self.register_hook(hook, priority='NORMAL')
+
+    def register_logger_hooks(self, log_config):
+        if log_config is None:
+            return
+        log_interval = log_config['interval']
+        for info in log_config['hooks']:
+            logger_hook = mmcv.build_from_cfg(
+                info, HOOKS, default_args=dict(interval=log_interval))
+            self.register_hook(logger_hook, priority='VERY_LOW')
+
+    def register_timer_hook(self, timer_config):
+        if timer_config is None:
+            return
+        if isinstance(timer_config, dict):
+            timer_config_ = copy.deepcopy(timer_config)
+            hook = mmcv.build_from_cfg(timer_config_, HOOKS)
+        else:
+            hook = timer_config
+        self.register_hook(hook, priority='LOW')
+
+    def register_custom_hooks(self, custom_config):
+        if custom_config is None:
+            return
+
+        if not isinstance(custom_config, list):
+            custom_config = [custom_config]
+
+        for item in custom_config:
+            if isinstance(item, dict):
+                self.register_hook_from_cfg(item)
+            else:
+                self.register_hook(item, priority='NORMAL')
+
+    def register_profiler_hook(self, profiler_config):
+        if profiler_config is None:
+            return
+        if isinstance(profiler_config, dict):
+            profiler_config.setdefault('type', 'ProfilerHook')
+            hook = mmcv.build_from_cfg(profiler_config, HOOKS)
+        else:
+            hook = profiler_config
+        self.register_hook(hook)
+
+    def register_training_hooks(self,
+                                lr_config,
+                                optimizer_config=None,
+                                checkpoint_config=None,
+                                log_config=None,
+                                momentum_config=None,
+                                timer_config=dict(type='IterTimerHook'),
+                                custom_hooks_config=None):
+        """Register default and custom hooks for training.
+
+        Default and custom hooks include:
+
+        +----------------------+-------------------------+
+        | Hooks                | Priority                |
+        +======================+=========================+
+        | LrUpdaterHook        | VERY_HIGH (10)          |
+        +----------------------+-------------------------+
+        | MomentumUpdaterHook  | HIGH (30)               |
+        +----------------------+-------------------------+
+        | OptimizerStepperHook | ABOVE_NORMAL (40)       |
+        +----------------------+-------------------------+
+        | CheckpointSaverHook  | NORMAL (50)             |
+        +----------------------+-------------------------+
+        | IterTimerHook        | LOW (70)                |
+        +----------------------+-------------------------+
+        | LoggerHook(s)        | VERY_LOW (90)           |
+        +----------------------+-------------------------+
+        | CustomHook(s)        | defaults to NORMAL (50) |
+        +----------------------+-------------------------+
+
+        If custom hooks have same priority with default hooks, custom hooks
+        will be triggered after default hooks.
+        """
+        self.register_lr_hook(lr_config)
+        self.register_momentum_hook(momentum_config)
+        self.register_optimizer_hook(optimizer_config)
+        self.register_checkpoint_hook(checkpoint_config)
+        self.register_timer_hook(timer_config)
+        self.register_logger_hooks(log_config)
+        self.register_custom_hooks(custom_hooks_config)
diff --git a/annotator/uniformer/mmcv/runner/builder.py b/annotator/uniformer/mmcv/runner/builder.py
new file mode 100644
index 0000000000000000000000000000000000000000..77c96ba0b2f30ead9da23f293c5dc84dd3e4a74f
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/builder.py
@@ -0,0 +1,24 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import copy
+
+from ..utils import Registry
+
+RUNNERS = Registry('runner')
+RUNNER_BUILDERS = Registry('runner builder')
+
+
+def build_runner_constructor(cfg):
+    return RUNNER_BUILDERS.build(cfg)
+
+
+def build_runner(cfg, default_args=None):
+    runner_cfg = copy.deepcopy(cfg)
+    constructor_type = runner_cfg.pop('constructor',
+                                      'DefaultRunnerConstructor')
+    runner_constructor = build_runner_constructor(
+        dict(
+            type=constructor_type,
+            runner_cfg=runner_cfg,
+            default_args=default_args))
+    runner = runner_constructor()
+    return runner
diff --git a/annotator/uniformer/mmcv/runner/checkpoint.py b/annotator/uniformer/mmcv/runner/checkpoint.py
new file mode 100644
index 0000000000000000000000000000000000000000..b29ca320679164432f446adad893e33fb2b4b29e
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/checkpoint.py
@@ -0,0 +1,707 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import io
+import os
+import os.path as osp
+import pkgutil
+import re
+import time
+import warnings
+from collections import OrderedDict
+from importlib import import_module
+from tempfile import TemporaryDirectory
+
+import torch
+import torchvision
+from torch.optim import Optimizer
+from torch.utils import model_zoo
+
+import annotator.uniformer.mmcv as mmcv
+from ..fileio import FileClient
+from ..fileio import load as load_file
+from ..parallel import is_module_wrapper
+from ..utils import mkdir_or_exist
+from .dist_utils import get_dist_info
+
+ENV_MMCV_HOME = 'MMCV_HOME'
+ENV_XDG_CACHE_HOME = 'XDG_CACHE_HOME'
+DEFAULT_CACHE_DIR = '~/.cache'
+
+
+def _get_mmcv_home():
+    mmcv_home = os.path.expanduser(
+        os.getenv(
+            ENV_MMCV_HOME,
+            os.path.join(
+                os.getenv(ENV_XDG_CACHE_HOME, DEFAULT_CACHE_DIR), 'mmcv')))
+
+    mkdir_or_exist(mmcv_home)
+    return mmcv_home
+
+
+def load_state_dict(module, state_dict, strict=False, logger=None):
+    """Load state_dict to a module.
+
+    This method is modified from :meth:`torch.nn.Module.load_state_dict`.
+    Default value for ``strict`` is set to ``False`` and the message for
+    param mismatch will be shown even if strict is False.
+
+    Args:
+        module (Module): Module that receives the state_dict.
+        state_dict (OrderedDict): Weights.
+        strict (bool): whether to strictly enforce that the keys
+            in :attr:`state_dict` match the keys returned by this module's
+            :meth:`~torch.nn.Module.state_dict` function. Default: ``False``.
+        logger (:obj:`logging.Logger`, optional): Logger to log the error
+            message. If not specified, print function will be used.
+    """
+    unexpected_keys = []
+    all_missing_keys = []
+    err_msg = []
+
+    metadata = getattr(state_dict, '_metadata', None)
+    state_dict = state_dict.copy()
+    if metadata is not None:
+        state_dict._metadata = metadata
+
+    # use _load_from_state_dict to enable checkpoint version control
+    def load(module, prefix=''):
+        # recursively check parallel module in case that the model has a
+        # complicated structure, e.g., nn.Module(nn.Module(DDP))
+        if is_module_wrapper(module):
+            module = module.module
+        local_metadata = {} if metadata is None else metadata.get(
+            prefix[:-1], {})
+        module._load_from_state_dict(state_dict, prefix, local_metadata, True,
+                                     all_missing_keys, unexpected_keys,
+                                     err_msg)
+        for name, child in module._modules.items():
+            if child is not None:
+                load(child, prefix + name + '.')
+
+    load(module)
+    load = None  # break load->load reference cycle
+
+    # ignore "num_batches_tracked" of BN layers
+    missing_keys = [
+        key for key in all_missing_keys if 'num_batches_tracked' not in key
+    ]
+
+    if unexpected_keys:
+        err_msg.append('unexpected key in source '
+                       f'state_dict: {", ".join(unexpected_keys)}\n')
+    if missing_keys:
+        err_msg.append(
+            f'missing keys in source state_dict: {", ".join(missing_keys)}\n')
+
+    rank, _ = get_dist_info()
+    if len(err_msg) > 0 and rank == 0:
+        err_msg.insert(
+            0, 'The model and loaded state dict do not match exactly\n')
+        err_msg = '\n'.join(err_msg)
+        if strict:
+            raise RuntimeError(err_msg)
+        elif logger is not None:
+            logger.warning(err_msg)
+        else:
+            print(err_msg)
+
+
+def get_torchvision_models():
+    model_urls = dict()
+    for _, name, ispkg in pkgutil.walk_packages(torchvision.models.__path__):
+        if ispkg:
+            continue
+        _zoo = import_module(f'torchvision.models.{name}')
+        if hasattr(_zoo, 'model_urls'):
+            _urls = getattr(_zoo, 'model_urls')
+            model_urls.update(_urls)
+    return model_urls
+
+
+def get_external_models():
+    mmcv_home = _get_mmcv_home()
+    default_json_path = osp.join(mmcv.__path__[0], 'model_zoo/open_mmlab.json')
+    default_urls = load_file(default_json_path)
+    assert isinstance(default_urls, dict)
+    external_json_path = osp.join(mmcv_home, 'open_mmlab.json')
+    if osp.exists(external_json_path):
+        external_urls = load_file(external_json_path)
+        assert isinstance(external_urls, dict)
+        default_urls.update(external_urls)
+
+    return default_urls
+
+
+def get_mmcls_models():
+    mmcls_json_path = osp.join(mmcv.__path__[0], 'model_zoo/mmcls.json')
+    mmcls_urls = load_file(mmcls_json_path)
+
+    return mmcls_urls
+
+
+def get_deprecated_model_names():
+    deprecate_json_path = osp.join(mmcv.__path__[0],
+                                   'model_zoo/deprecated.json')
+    deprecate_urls = load_file(deprecate_json_path)
+    assert isinstance(deprecate_urls, dict)
+
+    return deprecate_urls
+
+
+def _process_mmcls_checkpoint(checkpoint):
+    state_dict = checkpoint['state_dict']
+    new_state_dict = OrderedDict()
+    for k, v in state_dict.items():
+        if k.startswith('backbone.'):
+            new_state_dict[k[9:]] = v
+    new_checkpoint = dict(state_dict=new_state_dict)
+
+    return new_checkpoint
+
+
+class CheckpointLoader:
+    """A general checkpoint loader to manage all schemes."""
+
+    _schemes = {}
+
+    @classmethod
+    def _register_scheme(cls, prefixes, loader, force=False):
+        if isinstance(prefixes, str):
+            prefixes = [prefixes]
+        else:
+            assert isinstance(prefixes, (list, tuple))
+        for prefix in prefixes:
+            if (prefix not in cls._schemes) or force:
+                cls._schemes[prefix] = loader
+            else:
+                raise KeyError(
+                    f'{prefix} is already registered as a loader backend, '
+                    'add "force=True" if you want to override it')
+        # sort, longer prefixes take priority
+        cls._schemes = OrderedDict(
+            sorted(cls._schemes.items(), key=lambda t: t[0], reverse=True))
+
+    @classmethod
+    def register_scheme(cls, prefixes, loader=None, force=False):
+        """Register a loader to CheckpointLoader.
+
+        This method can be used as a normal class method or a decorator.
+
+        Args:
+            prefixes (str or list[str] or tuple[str]):
+            The prefix of the registered loader.
+            loader (function, optional): The loader function to be registered.
+                When this method is used as a decorator, loader is None.
+                Defaults to None.
+            force (bool, optional): Whether to override the loader
+                if the prefix has already been registered. Defaults to False.
+        """
+
+        if loader is not None:
+            cls._register_scheme(prefixes, loader, force=force)
+            return
+
+        def _register(loader_cls):
+            cls._register_scheme(prefixes, loader_cls, force=force)
+            return loader_cls
+
+        return _register
+
+    @classmethod
+    def _get_checkpoint_loader(cls, path):
+        """Finds a loader that supports the given path. Falls back to the local
+        loader if no other loader is found.
+
+        Args:
+            path (str): checkpoint path
+
+        Returns:
+            loader (function): checkpoint loader
+        """
+
+        for p in cls._schemes:
+            if path.startswith(p):
+                return cls._schemes[p]
+
+    @classmethod
+    def load_checkpoint(cls, filename, map_location=None, logger=None):
+        """load checkpoint through URL scheme path.
+
+        Args:
+            filename (str): checkpoint file name with given prefix
+            map_location (str, optional): Same as :func:`torch.load`.
+                Default: None
+            logger (:mod:`logging.Logger`, optional): The logger for message.
+                Default: None
+
+        Returns:
+            dict or OrderedDict: The loaded checkpoint.
+        """
+
+        checkpoint_loader = cls._get_checkpoint_loader(filename)
+        class_name = checkpoint_loader.__name__
+        mmcv.print_log(
+            f'load checkpoint from {class_name[10:]} path: {filename}', logger)
+        return checkpoint_loader(filename, map_location)
+
+
+@CheckpointLoader.register_scheme(prefixes='')
+def load_from_local(filename, map_location):
+    """load checkpoint by local file path.
+
+    Args:
+        filename (str): local checkpoint file path
+        map_location (str, optional): Same as :func:`torch.load`.
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+
+    if not osp.isfile(filename):
+        raise IOError(f'{filename} is not a checkpoint file')
+    checkpoint = torch.load(filename, map_location=map_location)
+    return checkpoint
+
+
+@CheckpointLoader.register_scheme(prefixes=('http://', 'https://'))
+def load_from_http(filename, map_location=None, model_dir=None):
+    """load checkpoint through HTTP or HTTPS scheme path. In distributed
+    setting, this function only download checkpoint at local rank 0.
+
+    Args:
+        filename (str): checkpoint file path with modelzoo or
+            torchvision prefix
+        map_location (str, optional): Same as :func:`torch.load`.
+        model_dir (string, optional): directory in which to save the object,
+            Default: None
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+    rank, world_size = get_dist_info()
+    rank = int(os.environ.get('LOCAL_RANK', rank))
+    if rank == 0:
+        checkpoint = model_zoo.load_url(
+            filename, model_dir=model_dir, map_location=map_location)
+    if world_size > 1:
+        torch.distributed.barrier()
+        if rank > 0:
+            checkpoint = model_zoo.load_url(
+                filename, model_dir=model_dir, map_location=map_location)
+    return checkpoint
+
+
+@CheckpointLoader.register_scheme(prefixes='pavi://')
+def load_from_pavi(filename, map_location=None):
+    """load checkpoint through the file path prefixed with pavi. In distributed
+    setting, this function download ckpt at all ranks to different temporary
+    directories.
+
+    Args:
+        filename (str): checkpoint file path with pavi prefix
+        map_location (str, optional): Same as :func:`torch.load`.
+          Default: None
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+    assert filename.startswith('pavi://'), \
+        f'Expected filename startswith `pavi://`, but get {filename}'
+    model_path = filename[7:]
+
+    try:
+        from pavi import modelcloud
+    except ImportError:
+        raise ImportError(
+            'Please install pavi to load checkpoint from modelcloud.')
+
+    model = modelcloud.get(model_path)
+    with TemporaryDirectory() as tmp_dir:
+        downloaded_file = osp.join(tmp_dir, model.name)
+        model.download(downloaded_file)
+        checkpoint = torch.load(downloaded_file, map_location=map_location)
+    return checkpoint
+
+
+@CheckpointLoader.register_scheme(prefixes='s3://')
+def load_from_ceph(filename, map_location=None, backend='petrel'):
+    """load checkpoint through the file path prefixed with s3.  In distributed
+    setting, this function download ckpt at all ranks to different temporary
+    directories.
+
+    Args:
+        filename (str): checkpoint file path with s3 prefix
+        map_location (str, optional): Same as :func:`torch.load`.
+        backend (str, optional): The storage backend type. Options are 'ceph',
+            'petrel'. Default: 'petrel'.
+
+    .. warning::
+        :class:`mmcv.fileio.file_client.CephBackend` will be deprecated,
+        please use :class:`mmcv.fileio.file_client.PetrelBackend` instead.
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+    allowed_backends = ['ceph', 'petrel']
+    if backend not in allowed_backends:
+        raise ValueError(f'Load from Backend {backend} is not supported.')
+
+    if backend == 'ceph':
+        warnings.warn(
+            'CephBackend will be deprecated, please use PetrelBackend instead')
+
+    # CephClient and PetrelBackend have the same prefix 's3://' and the latter
+    # will be chosen as default. If PetrelBackend can not be instantiated
+    # successfully, the CephClient will be chosen.
+    try:
+        file_client = FileClient(backend=backend)
+    except ImportError:
+        allowed_backends.remove(backend)
+        file_client = FileClient(backend=allowed_backends[0])
+
+    with io.BytesIO(file_client.get(filename)) as buffer:
+        checkpoint = torch.load(buffer, map_location=map_location)
+    return checkpoint
+
+
+@CheckpointLoader.register_scheme(prefixes=('modelzoo://', 'torchvision://'))
+def load_from_torchvision(filename, map_location=None):
+    """load checkpoint through the file path prefixed with modelzoo or
+    torchvision.
+
+    Args:
+        filename (str): checkpoint file path with modelzoo or
+            torchvision prefix
+        map_location (str, optional): Same as :func:`torch.load`.
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+    model_urls = get_torchvision_models()
+    if filename.startswith('modelzoo://'):
+        warnings.warn('The URL scheme of "modelzoo://" is deprecated, please '
+                      'use "torchvision://" instead')
+        model_name = filename[11:]
+    else:
+        model_name = filename[14:]
+    return load_from_http(model_urls[model_name], map_location=map_location)
+
+
+@CheckpointLoader.register_scheme(prefixes=('open-mmlab://', 'openmmlab://'))
+def load_from_openmmlab(filename, map_location=None):
+    """load checkpoint through the file path prefixed with open-mmlab or
+    openmmlab.
+
+    Args:
+        filename (str): checkpoint file path with open-mmlab or
+        openmmlab prefix
+        map_location (str, optional): Same as :func:`torch.load`.
+          Default: None
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+
+    model_urls = get_external_models()
+    prefix_str = 'open-mmlab://'
+    if filename.startswith(prefix_str):
+        model_name = filename[13:]
+    else:
+        model_name = filename[12:]
+        prefix_str = 'openmmlab://'
+
+    deprecated_urls = get_deprecated_model_names()
+    if model_name in deprecated_urls:
+        warnings.warn(f'{prefix_str}{model_name} is deprecated in favor '
+                      f'of {prefix_str}{deprecated_urls[model_name]}')
+        model_name = deprecated_urls[model_name]
+    model_url = model_urls[model_name]
+    # check if is url
+    if model_url.startswith(('http://', 'https://')):
+        checkpoint = load_from_http(model_url, map_location=map_location)
+    else:
+        filename = osp.join(_get_mmcv_home(), model_url)
+        if not osp.isfile(filename):
+            raise IOError(f'{filename} is not a checkpoint file')
+        checkpoint = torch.load(filename, map_location=map_location)
+    return checkpoint
+
+
+@CheckpointLoader.register_scheme(prefixes='mmcls://')
+def load_from_mmcls(filename, map_location=None):
+    """load checkpoint through the file path prefixed with mmcls.
+
+    Args:
+        filename (str): checkpoint file path with mmcls prefix
+        map_location (str, optional): Same as :func:`torch.load`.
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+
+    model_urls = get_mmcls_models()
+    model_name = filename[8:]
+    checkpoint = load_from_http(
+        model_urls[model_name], map_location=map_location)
+    checkpoint = _process_mmcls_checkpoint(checkpoint)
+    return checkpoint
+
+
+def _load_checkpoint(filename, map_location=None, logger=None):
+    """Load checkpoint from somewhere (modelzoo, file, url).
+
+    Args:
+        filename (str): Accept local filepath, URL, ``torchvision://xxx``,
+            ``open-mmlab://xxx``. Please refer to ``docs/model_zoo.md`` for
+            details.
+        map_location (str, optional): Same as :func:`torch.load`.
+           Default: None.
+        logger (:mod:`logging.Logger`, optional): The logger for error message.
+           Default: None
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint. It can be either an
+           OrderedDict storing model weights or a dict containing other
+           information, which depends on the checkpoint.
+    """
+    return CheckpointLoader.load_checkpoint(filename, map_location, logger)
+
+
+def _load_checkpoint_with_prefix(prefix, filename, map_location=None):
+    """Load partial pretrained model with specific prefix.
+
+    Args:
+        prefix (str): The prefix of sub-module.
+        filename (str): Accept local filepath, URL, ``torchvision://xxx``,
+            ``open-mmlab://xxx``. Please refer to ``docs/model_zoo.md`` for
+            details.
+        map_location (str | None): Same as :func:`torch.load`. Default: None.
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+
+    checkpoint = _load_checkpoint(filename, map_location=map_location)
+
+    if 'state_dict' in checkpoint:
+        state_dict = checkpoint['state_dict']
+    else:
+        state_dict = checkpoint
+    if not prefix.endswith('.'):
+        prefix += '.'
+    prefix_len = len(prefix)
+
+    state_dict = {
+        k[prefix_len:]: v
+        for k, v in state_dict.items() if k.startswith(prefix)
+    }
+
+    assert state_dict, f'{prefix} is not in the pretrained model'
+    return state_dict
+
+
+def load_checkpoint(model,
+                    filename,
+                    map_location=None,
+                    strict=False,
+                    logger=None,
+                    revise_keys=[(r'^module\.', '')]):
+    """Load checkpoint from a file or URI.
+
+    Args:
+        model (Module): Module to load checkpoint.
+        filename (str): Accept local filepath, URL, ``torchvision://xxx``,
+            ``open-mmlab://xxx``. Please refer to ``docs/model_zoo.md`` for
+            details.
+        map_location (str): Same as :func:`torch.load`.
+        strict (bool): Whether to allow different params for the model and
+            checkpoint.
+        logger (:mod:`logging.Logger` or None): The logger for error message.
+        revise_keys (list): A list of customized keywords to modify the
+            state_dict in checkpoint. Each item is a (pattern, replacement)
+            pair of the regular expression operations. Default: strip
+            the prefix 'module.' by [(r'^module\\.', '')].
+
+    Returns:
+        dict or OrderedDict: The loaded checkpoint.
+    """
+    checkpoint = _load_checkpoint(filename, map_location, logger)
+    # OrderedDict is a subclass of dict
+    if not isinstance(checkpoint, dict):
+        raise RuntimeError(
+            f'No state_dict found in checkpoint file {filename}')
+    # get state_dict from checkpoint
+    if 'state_dict' in checkpoint:
+        state_dict = checkpoint['state_dict']
+    else:
+        state_dict = checkpoint
+
+    # strip prefix of state_dict
+    metadata = getattr(state_dict, '_metadata', OrderedDict())
+    for p, r in revise_keys:
+        state_dict = OrderedDict(
+            {re.sub(p, r, k): v
+             for k, v in state_dict.items()})
+    # Keep metadata in state_dict
+    state_dict._metadata = metadata
+
+    # load state_dict
+    load_state_dict(model, state_dict, strict, logger)
+    return checkpoint
+
+
+def weights_to_cpu(state_dict):
+    """Copy a model state_dict to cpu.
+
+    Args:
+        state_dict (OrderedDict): Model weights on GPU.
+
+    Returns:
+        OrderedDict: Model weights on GPU.
+    """
+    state_dict_cpu = OrderedDict()
+    for key, val in state_dict.items():
+        state_dict_cpu[key] = val.cpu()
+    # Keep metadata in state_dict
+    state_dict_cpu._metadata = getattr(state_dict, '_metadata', OrderedDict())
+    return state_dict_cpu
+
+
+def _save_to_state_dict(module, destination, prefix, keep_vars):
+    """Saves module state to `destination` dictionary.
+
+    This method is modified from :meth:`torch.nn.Module._save_to_state_dict`.
+
+    Args:
+        module (nn.Module): The module to generate state_dict.
+        destination (dict): A dict where state will be stored.
+        prefix (str): The prefix for parameters and buffers used in this
+            module.
+    """
+    for name, param in module._parameters.items():
+        if param is not None:
+            destination[prefix + name] = param if keep_vars else param.detach()
+    for name, buf in module._buffers.items():
+        # remove check of _non_persistent_buffers_set to allow nn.BatchNorm2d
+        if buf is not None:
+            destination[prefix + name] = buf if keep_vars else buf.detach()
+
+
+def get_state_dict(module, destination=None, prefix='', keep_vars=False):
+    """Returns a dictionary containing a whole state of the module.
+
+    Both parameters and persistent buffers (e.g. running averages) are
+    included. Keys are corresponding parameter and buffer names.
+
+    This method is modified from :meth:`torch.nn.Module.state_dict` to
+    recursively check parallel module in case that the model has a complicated
+    structure, e.g., nn.Module(nn.Module(DDP)).
+
+    Args:
+        module (nn.Module): The module to generate state_dict.
+        destination (OrderedDict): Returned dict for the state of the
+            module.
+        prefix (str): Prefix of the key.
+        keep_vars (bool): Whether to keep the variable property of the
+            parameters. Default: False.
+
+    Returns:
+        dict: A dictionary containing a whole state of the module.
+    """
+    # recursively check parallel module in case that the model has a
+    # complicated structure, e.g., nn.Module(nn.Module(DDP))
+    if is_module_wrapper(module):
+        module = module.module
+
+    # below is the same as torch.nn.Module.state_dict()
+    if destination is None:
+        destination = OrderedDict()
+        destination._metadata = OrderedDict()
+    destination._metadata[prefix[:-1]] = local_metadata = dict(
+        version=module._version)
+    _save_to_state_dict(module, destination, prefix, keep_vars)
+    for name, child in module._modules.items():
+        if child is not None:
+            get_state_dict(
+                child, destination, prefix + name + '.', keep_vars=keep_vars)
+    for hook in module._state_dict_hooks.values():
+        hook_result = hook(module, destination, prefix, local_metadata)
+        if hook_result is not None:
+            destination = hook_result
+    return destination
+
+
+def save_checkpoint(model,
+                    filename,
+                    optimizer=None,
+                    meta=None,
+                    file_client_args=None):
+    """Save checkpoint to file.
+
+    The checkpoint will have 3 fields: ``meta``, ``state_dict`` and
+    ``optimizer``. By default ``meta`` will contain version and time info.
+
+    Args:
+        model (Module): Module whose params are to be saved.
+        filename (str): Checkpoint filename.
+        optimizer (:obj:`Optimizer`, optional): Optimizer to be saved.
+        meta (dict, optional): Metadata to be saved in checkpoint.
+        file_client_args (dict, optional): Arguments to instantiate a
+            FileClient. See :class:`mmcv.fileio.FileClient` for details.
+            Default: None.
+            `New in version 1.3.16.`
+    """
+    if meta is None:
+        meta = {}
+    elif not isinstance(meta, dict):
+        raise TypeError(f'meta must be a dict or None, but got {type(meta)}')
+    meta.update(mmcv_version=mmcv.__version__, time=time.asctime())
+
+    if is_module_wrapper(model):
+        model = model.module
+
+    if hasattr(model, 'CLASSES') and model.CLASSES is not None:
+        # save class name to the meta
+        meta.update(CLASSES=model.CLASSES)
+
+    checkpoint = {
+        'meta': meta,
+        'state_dict': weights_to_cpu(get_state_dict(model))
+    }
+    # save optimizer state dict in the checkpoint
+    if isinstance(optimizer, Optimizer):
+        checkpoint['optimizer'] = optimizer.state_dict()
+    elif isinstance(optimizer, dict):
+        checkpoint['optimizer'] = {}
+        for name, optim in optimizer.items():
+            checkpoint['optimizer'][name] = optim.state_dict()
+
+    if filename.startswith('pavi://'):
+        if file_client_args is not None:
+            raise ValueError(
+                'file_client_args should be "None" if filename starts with'
+                f'"pavi://", but got {file_client_args}')
+        try:
+            from pavi import modelcloud
+            from pavi import exception
+        except ImportError:
+            raise ImportError(
+                'Please install pavi to load checkpoint from modelcloud.')
+        model_path = filename[7:]
+        root = modelcloud.Folder()
+        model_dir, model_name = osp.split(model_path)
+        try:
+            model = modelcloud.get(model_dir)
+        except exception.NodeNotFoundError:
+            model = root.create_training_model(model_dir)
+        with TemporaryDirectory() as tmp_dir:
+            checkpoint_file = osp.join(tmp_dir, model_name)
+            with open(checkpoint_file, 'wb') as f:
+                torch.save(checkpoint, f)
+                f.flush()
+            model.create_file(checkpoint_file, name=model_name)
+    else:
+        file_client = FileClient.infer_client(file_client_args, filename)
+        with io.BytesIO() as f:
+            torch.save(checkpoint, f)
+            file_client.put(f.getvalue(), filename)
diff --git a/annotator/uniformer/mmcv/runner/default_constructor.py b/annotator/uniformer/mmcv/runner/default_constructor.py
new file mode 100644
index 0000000000000000000000000000000000000000..3f1f5b44168768dfda3947393a63a6cf9cf50b41
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/default_constructor.py
@@ -0,0 +1,44 @@
+from .builder import RUNNER_BUILDERS, RUNNERS
+
+
+@RUNNER_BUILDERS.register_module()
+class DefaultRunnerConstructor:
+    """Default constructor for runners.
+
+    Custom existing `Runner` like `EpocBasedRunner` though `RunnerConstructor`.
+    For example, We can inject some new properties and functions for `Runner`.
+
+    Example:
+        >>> from annotator.uniformer.mmcv.runner import RUNNER_BUILDERS, build_runner
+        >>> # Define a new RunnerReconstructor
+        >>> @RUNNER_BUILDERS.register_module()
+        >>> class MyRunnerConstructor:
+        ...     def __init__(self, runner_cfg, default_args=None):
+        ...         if not isinstance(runner_cfg, dict):
+        ...             raise TypeError('runner_cfg should be a dict',
+        ...                             f'but got {type(runner_cfg)}')
+        ...         self.runner_cfg = runner_cfg
+        ...         self.default_args = default_args
+        ...
+        ...     def __call__(self):
+        ...         runner = RUNNERS.build(self.runner_cfg,
+        ...                                default_args=self.default_args)
+        ...         # Add new properties for existing runner
+        ...         runner.my_name = 'my_runner'
+        ...         runner.my_function = lambda self: print(self.my_name)
+        ...         ...
+        >>> # build your runner
+        >>> runner_cfg = dict(type='EpochBasedRunner', max_epochs=40,
+        ...                   constructor='MyRunnerConstructor')
+        >>> runner = build_runner(runner_cfg)
+    """
+
+    def __init__(self, runner_cfg, default_args=None):
+        if not isinstance(runner_cfg, dict):
+            raise TypeError('runner_cfg should be a dict',
+                            f'but got {type(runner_cfg)}')
+        self.runner_cfg = runner_cfg
+        self.default_args = default_args
+
+    def __call__(self):
+        return RUNNERS.build(self.runner_cfg, default_args=self.default_args)
diff --git a/annotator/uniformer/mmcv/runner/dist_utils.py b/annotator/uniformer/mmcv/runner/dist_utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..d3a1ef3fda5ceeb31bf15a73779da1b1903ab0fe
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/dist_utils.py
@@ -0,0 +1,164 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import functools
+import os
+import subprocess
+from collections import OrderedDict
+
+import torch
+import torch.multiprocessing as mp
+from torch import distributed as dist
+from torch._utils import (_flatten_dense_tensors, _take_tensors,
+                          _unflatten_dense_tensors)
+
+
+def init_dist(launcher, backend='nccl', **kwargs):
+    if mp.get_start_method(allow_none=True) is None:
+        mp.set_start_method('spawn')
+    if launcher == 'pytorch':
+        _init_dist_pytorch(backend, **kwargs)
+    elif launcher == 'mpi':
+        _init_dist_mpi(backend, **kwargs)
+    elif launcher == 'slurm':
+        _init_dist_slurm(backend, **kwargs)
+    else:
+        raise ValueError(f'Invalid launcher type: {launcher}')
+
+
+def _init_dist_pytorch(backend, **kwargs):
+    # TODO: use local_rank instead of rank % num_gpus
+    rank = int(os.environ['RANK'])
+    num_gpus = torch.cuda.device_count()
+    torch.cuda.set_device(rank % num_gpus)
+    dist.init_process_group(backend=backend, **kwargs)
+
+
+def _init_dist_mpi(backend, **kwargs):
+    # TODO: use local_rank instead of rank % num_gpus
+    rank = int(os.environ['OMPI_COMM_WORLD_RANK'])
+    num_gpus = torch.cuda.device_count()
+    torch.cuda.set_device(rank % num_gpus)
+    dist.init_process_group(backend=backend, **kwargs)
+
+
+def _init_dist_slurm(backend, port=None):
+    """Initialize slurm distributed training environment.
+
+    If argument ``port`` is not specified, then the master port will be system
+    environment variable ``MASTER_PORT``. If ``MASTER_PORT`` is not in system
+    environment variable, then a default port ``29500`` will be used.
+
+    Args:
+        backend (str): Backend of torch.distributed.
+        port (int, optional): Master port. Defaults to None.
+    """
+    proc_id = int(os.environ['SLURM_PROCID'])
+    ntasks = int(os.environ['SLURM_NTASKS'])
+    node_list = os.environ['SLURM_NODELIST']
+    num_gpus = torch.cuda.device_count()
+    torch.cuda.set_device(proc_id % num_gpus)
+    addr = subprocess.getoutput(
+        f'scontrol show hostname {node_list} | head -n1')
+    # specify master port
+    if port is not None:
+        os.environ['MASTER_PORT'] = str(port)
+    elif 'MASTER_PORT' in os.environ:
+        pass  # use MASTER_PORT in the environment variable
+    else:
+        # 29500 is torch.distributed default port
+        os.environ['MASTER_PORT'] = '29500'
+    # use MASTER_ADDR in the environment variable if it already exists
+    if 'MASTER_ADDR' not in os.environ:
+        os.environ['MASTER_ADDR'] = addr
+    os.environ['WORLD_SIZE'] = str(ntasks)
+    os.environ['LOCAL_RANK'] = str(proc_id % num_gpus)
+    os.environ['RANK'] = str(proc_id)
+    dist.init_process_group(backend=backend)
+
+
+def get_dist_info():
+    if dist.is_available() and dist.is_initialized():
+        rank = dist.get_rank()
+        world_size = dist.get_world_size()
+    else:
+        rank = 0
+        world_size = 1
+    return rank, world_size
+
+
+def master_only(func):
+
+    @functools.wraps(func)
+    def wrapper(*args, **kwargs):
+        rank, _ = get_dist_info()
+        if rank == 0:
+            return func(*args, **kwargs)
+
+    return wrapper
+
+
+def allreduce_params(params, coalesce=True, bucket_size_mb=-1):
+    """Allreduce parameters.
+
+    Args:
+        params (list[torch.Parameters]): List of parameters or buffers of a
+            model.
+        coalesce (bool, optional): Whether allreduce parameters as a whole.
+            Defaults to True.
+        bucket_size_mb (int, optional): Size of bucket, the unit is MB.
+            Defaults to -1.
+    """
+    _, world_size = get_dist_info()
+    if world_size == 1:
+        return
+    params = [param.data for param in params]
+    if coalesce:
+        _allreduce_coalesced(params, world_size, bucket_size_mb)
+    else:
+        for tensor in params:
+            dist.all_reduce(tensor.div_(world_size))
+
+
+def allreduce_grads(params, coalesce=True, bucket_size_mb=-1):
+    """Allreduce gradients.
+
+    Args:
+        params (list[torch.Parameters]): List of parameters of a model
+        coalesce (bool, optional): Whether allreduce parameters as a whole.
+            Defaults to True.
+        bucket_size_mb (int, optional): Size of bucket, the unit is MB.
+            Defaults to -1.
+    """
+    grads = [
+        param.grad.data for param in params
+        if param.requires_grad and param.grad is not None
+    ]
+    _, world_size = get_dist_info()
+    if world_size == 1:
+        return
+    if coalesce:
+        _allreduce_coalesced(grads, world_size, bucket_size_mb)
+    else:
+        for tensor in grads:
+            dist.all_reduce(tensor.div_(world_size))
+
+
+def _allreduce_coalesced(tensors, world_size, bucket_size_mb=-1):
+    if bucket_size_mb > 0:
+        bucket_size_bytes = bucket_size_mb * 1024 * 1024
+        buckets = _take_tensors(tensors, bucket_size_bytes)
+    else:
+        buckets = OrderedDict()
+        for tensor in tensors:
+            tp = tensor.type()
+            if tp not in buckets:
+                buckets[tp] = []
+            buckets[tp].append(tensor)
+        buckets = buckets.values()
+
+    for bucket in buckets:
+        flat_tensors = _flatten_dense_tensors(bucket)
+        dist.all_reduce(flat_tensors)
+        flat_tensors.div_(world_size)
+        for tensor, synced in zip(
+                bucket, _unflatten_dense_tensors(flat_tensors, bucket)):
+            tensor.copy_(synced)
diff --git a/annotator/uniformer/mmcv/runner/epoch_based_runner.py b/annotator/uniformer/mmcv/runner/epoch_based_runner.py
new file mode 100644
index 0000000000000000000000000000000000000000..766a9ce6afdf09cd11b1b15005f5132583011348
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/epoch_based_runner.py
@@ -0,0 +1,187 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import os.path as osp
+import platform
+import shutil
+import time
+import warnings
+
+import torch
+
+import annotator.uniformer.mmcv as mmcv
+from .base_runner import BaseRunner
+from .builder import RUNNERS
+from .checkpoint import save_checkpoint
+from .utils import get_host_info
+
+
+@RUNNERS.register_module()
+class EpochBasedRunner(BaseRunner):
+    """Epoch-based Runner.
+
+    This runner train models epoch by epoch.
+    """
+
+    def run_iter(self, data_batch, train_mode, **kwargs):
+        if self.batch_processor is not None:
+            outputs = self.batch_processor(
+                self.model, data_batch, train_mode=train_mode, **kwargs)
+        elif train_mode:
+            outputs = self.model.train_step(data_batch, self.optimizer,
+                                            **kwargs)
+        else:
+            outputs = self.model.val_step(data_batch, self.optimizer, **kwargs)
+        if not isinstance(outputs, dict):
+            raise TypeError('"batch_processor()" or "model.train_step()"'
+                            'and "model.val_step()" must return a dict')
+        if 'log_vars' in outputs:
+            self.log_buffer.update(outputs['log_vars'], outputs['num_samples'])
+        self.outputs = outputs
+
+    def train(self, data_loader, **kwargs):
+        self.model.train()
+        self.mode = 'train'
+        self.data_loader = data_loader
+        self._max_iters = self._max_epochs * len(self.data_loader)
+        self.call_hook('before_train_epoch')
+        time.sleep(2)  # Prevent possible deadlock during epoch transition
+        for i, data_batch in enumerate(self.data_loader):
+            self._inner_iter = i
+            self.call_hook('before_train_iter')
+            self.run_iter(data_batch, train_mode=True, **kwargs)
+            self.call_hook('after_train_iter')
+            self._iter += 1
+
+        self.call_hook('after_train_epoch')
+        self._epoch += 1
+
+    @torch.no_grad()
+    def val(self, data_loader, **kwargs):
+        self.model.eval()
+        self.mode = 'val'
+        self.data_loader = data_loader
+        self.call_hook('before_val_epoch')
+        time.sleep(2)  # Prevent possible deadlock during epoch transition
+        for i, data_batch in enumerate(self.data_loader):
+            self._inner_iter = i
+            self.call_hook('before_val_iter')
+            self.run_iter(data_batch, train_mode=False)
+            self.call_hook('after_val_iter')
+
+        self.call_hook('after_val_epoch')
+
+    def run(self, data_loaders, workflow, max_epochs=None, **kwargs):
+        """Start running.
+
+        Args:
+            data_loaders (list[:obj:`DataLoader`]): Dataloaders for training
+                and validation.
+            workflow (list[tuple]): A list of (phase, epochs) to specify the
+                running order and epochs. E.g, [('train', 2), ('val', 1)] means
+                running 2 epochs for training and 1 epoch for validation,
+                iteratively.
+        """
+        assert isinstance(data_loaders, list)
+        assert mmcv.is_list_of(workflow, tuple)
+        assert len(data_loaders) == len(workflow)
+        if max_epochs is not None:
+            warnings.warn(
+                'setting max_epochs in run is deprecated, '
+                'please set max_epochs in runner_config', DeprecationWarning)
+            self._max_epochs = max_epochs
+
+        assert self._max_epochs is not None, (
+            'max_epochs must be specified during instantiation')
+
+        for i, flow in enumerate(workflow):
+            mode, epochs = flow
+            if mode == 'train':
+                self._max_iters = self._max_epochs * len(data_loaders[i])
+                break
+
+        work_dir = self.work_dir if self.work_dir is not None else 'NONE'
+        self.logger.info('Start running, host: %s, work_dir: %s',
+                         get_host_info(), work_dir)
+        self.logger.info('Hooks will be executed in the following order:\n%s',
+                         self.get_hook_info())
+        self.logger.info('workflow: %s, max: %d epochs', workflow,
+                         self._max_epochs)
+        self.call_hook('before_run')
+
+        while self.epoch < self._max_epochs:
+            for i, flow in enumerate(workflow):
+                mode, epochs = flow
+                if isinstance(mode, str):  # self.train()
+                    if not hasattr(self, mode):
+                        raise ValueError(
+                            f'runner has no method named "{mode}" to run an '
+                            'epoch')
+                    epoch_runner = getattr(self, mode)
+                else:
+                    raise TypeError(
+                        'mode in workflow must be a str, but got {}'.format(
+                            type(mode)))
+
+                for _ in range(epochs):
+                    if mode == 'train' and self.epoch >= self._max_epochs:
+                        break
+                    epoch_runner(data_loaders[i], **kwargs)
+
+        time.sleep(1)  # wait for some hooks like loggers to finish
+        self.call_hook('after_run')
+
+    def save_checkpoint(self,
+                        out_dir,
+                        filename_tmpl='epoch_{}.pth',
+                        save_optimizer=True,
+                        meta=None,
+                        create_symlink=True):
+        """Save the checkpoint.
+
+        Args:
+            out_dir (str): The directory that checkpoints are saved.
+            filename_tmpl (str, optional): The checkpoint filename template,
+                which contains a placeholder for the epoch number.
+                Defaults to 'epoch_{}.pth'.
+            save_optimizer (bool, optional): Whether to save the optimizer to
+                the checkpoint. Defaults to True.
+            meta (dict, optional): The meta information to be saved in the
+                checkpoint. Defaults to None.
+            create_symlink (bool, optional): Whether to create a symlink
+                "latest.pth" to point to the latest checkpoint.
+                Defaults to True.
+        """
+        if meta is None:
+            meta = {}
+        elif not isinstance(meta, dict):
+            raise TypeError(
+                f'meta should be a dict or None, but got {type(meta)}')
+        if self.meta is not None:
+            meta.update(self.meta)
+            # Note: meta.update(self.meta) should be done before
+            # meta.update(epoch=self.epoch + 1, iter=self.iter) otherwise
+            # there will be problems with resumed checkpoints.
+            # More details in https://github.com/open-mmlab/mmcv/pull/1108
+        meta.update(epoch=self.epoch + 1, iter=self.iter)
+
+        filename = filename_tmpl.format(self.epoch + 1)
+        filepath = osp.join(out_dir, filename)
+        optimizer = self.optimizer if save_optimizer else None
+        save_checkpoint(self.model, filepath, optimizer=optimizer, meta=meta)
+        # in some environments, `os.symlink` is not supported, you may need to
+        # set `create_symlink` to False
+        if create_symlink:
+            dst_file = osp.join(out_dir, 'latest.pth')
+            if platform.system() != 'Windows':
+                mmcv.symlink(filename, dst_file)
+            else:
+                shutil.copy(filepath, dst_file)
+
+
+@RUNNERS.register_module()
+class Runner(EpochBasedRunner):
+    """Deprecated name of EpochBasedRunner."""
+
+    def __init__(self, *args, **kwargs):
+        warnings.warn(
+            'Runner was deprecated, please use EpochBasedRunner instead')
+        super().__init__(*args, **kwargs)
diff --git a/annotator/uniformer/mmcv/runner/fp16_utils.py b/annotator/uniformer/mmcv/runner/fp16_utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..1981011d6859192e3e663e29d13500d56ba47f6c
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/fp16_utils.py
@@ -0,0 +1,410 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import functools
+import warnings
+from collections import abc
+from inspect import getfullargspec
+
+import numpy as np
+import torch
+import torch.nn as nn
+
+from annotator.uniformer.mmcv.utils import TORCH_VERSION, digit_version
+from .dist_utils import allreduce_grads as _allreduce_grads
+
+try:
+    # If PyTorch version >= 1.6.0, torch.cuda.amp.autocast would be imported
+    # and used; otherwise, auto fp16 will adopt mmcv's implementation.
+    # Note that when PyTorch >= 1.6.0, we still cast tensor types to fp16
+    # manually, so the behavior may not be consistent with real amp.
+    from torch.cuda.amp import autocast
+except ImportError:
+    pass
+
+
+def cast_tensor_type(inputs, src_type, dst_type):
+    """Recursively convert Tensor in inputs from src_type to dst_type.
+
+    Args:
+        inputs: Inputs that to be casted.
+        src_type (torch.dtype): Source type..
+        dst_type (torch.dtype): Destination type.
+
+    Returns:
+        The same type with inputs, but all contained Tensors have been cast.
+    """
+    if isinstance(inputs, nn.Module):
+        return inputs
+    elif isinstance(inputs, torch.Tensor):
+        return inputs.to(dst_type)
+    elif isinstance(inputs, str):
+        return inputs
+    elif isinstance(inputs, np.ndarray):
+        return inputs
+    elif isinstance(inputs, abc.Mapping):
+        return type(inputs)({
+            k: cast_tensor_type(v, src_type, dst_type)
+            for k, v in inputs.items()
+        })
+    elif isinstance(inputs, abc.Iterable):
+        return type(inputs)(
+            cast_tensor_type(item, src_type, dst_type) for item in inputs)
+    else:
+        return inputs
+
+
+def auto_fp16(apply_to=None, out_fp32=False):
+    """Decorator to enable fp16 training automatically.
+
+    This decorator is useful when you write custom modules and want to support
+    mixed precision training. If inputs arguments are fp32 tensors, they will
+    be converted to fp16 automatically. Arguments other than fp32 tensors are
+    ignored. If you are using PyTorch >= 1.6, torch.cuda.amp is used as the
+    backend, otherwise, original mmcv implementation will be adopted.
+
+    Args:
+        apply_to (Iterable, optional): The argument names to be converted.
+            `None` indicates all arguments.
+        out_fp32 (bool): Whether to convert the output back to fp32.
+
+    Example:
+
+        >>> import torch.nn as nn
+        >>> class MyModule1(nn.Module):
+        >>>
+        >>>     # Convert x and y to fp16
+        >>>     @auto_fp16()
+        >>>     def forward(self, x, y):
+        >>>         pass
+
+        >>> import torch.nn as nn
+        >>> class MyModule2(nn.Module):
+        >>>
+        >>>     # convert pred to fp16
+        >>>     @auto_fp16(apply_to=('pred', ))
+        >>>     def do_something(self, pred, others):
+        >>>         pass
+    """
+
+    def auto_fp16_wrapper(old_func):
+
+        @functools.wraps(old_func)
+        def new_func(*args, **kwargs):
+            # check if the module has set the attribute `fp16_enabled`, if not,
+            # just fallback to the original method.
+            if not isinstance(args[0], torch.nn.Module):
+                raise TypeError('@auto_fp16 can only be used to decorate the '
+                                'method of nn.Module')
+            if not (hasattr(args[0], 'fp16_enabled') and args[0].fp16_enabled):
+                return old_func(*args, **kwargs)
+
+            # get the arg spec of the decorated method
+            args_info = getfullargspec(old_func)
+            # get the argument names to be casted
+            args_to_cast = args_info.args if apply_to is None else apply_to
+            # convert the args that need to be processed
+            new_args = []
+            # NOTE: default args are not taken into consideration
+            if args:
+                arg_names = args_info.args[:len(args)]
+                for i, arg_name in enumerate(arg_names):
+                    if arg_name in args_to_cast:
+                        new_args.append(
+                            cast_tensor_type(args[i], torch.float, torch.half))
+                    else:
+                        new_args.append(args[i])
+            # convert the kwargs that need to be processed
+            new_kwargs = {}
+            if kwargs:
+                for arg_name, arg_value in kwargs.items():
+                    if arg_name in args_to_cast:
+                        new_kwargs[arg_name] = cast_tensor_type(
+                            arg_value, torch.float, torch.half)
+                    else:
+                        new_kwargs[arg_name] = arg_value
+            # apply converted arguments to the decorated method
+            if (TORCH_VERSION != 'parrots' and
+                    digit_version(TORCH_VERSION) >= digit_version('1.6.0')):
+                with autocast(enabled=True):
+                    output = old_func(*new_args, **new_kwargs)
+            else:
+                output = old_func(*new_args, **new_kwargs)
+            # cast the results back to fp32 if necessary
+            if out_fp32:
+                output = cast_tensor_type(output, torch.half, torch.float)
+            return output
+
+        return new_func
+
+    return auto_fp16_wrapper
+
+
+def force_fp32(apply_to=None, out_fp16=False):
+    """Decorator to convert input arguments to fp32 in force.
+
+    This decorator is useful when you write custom modules and want to support
+    mixed precision training. If there are some inputs that must be processed
+    in fp32 mode, then this decorator can handle it. If inputs arguments are
+    fp16 tensors, they will be converted to fp32 automatically. Arguments other
+    than fp16 tensors are ignored. If you are using PyTorch >= 1.6,
+    torch.cuda.amp is used as the backend, otherwise, original mmcv
+    implementation will be adopted.
+
+    Args:
+        apply_to (Iterable, optional): The argument names to be converted.
+            `None` indicates all arguments.
+        out_fp16 (bool): Whether to convert the output back to fp16.
+
+    Example:
+
+        >>> import torch.nn as nn
+        >>> class MyModule1(nn.Module):
+        >>>
+        >>>     # Convert x and y to fp32
+        >>>     @force_fp32()
+        >>>     def loss(self, x, y):
+        >>>         pass
+
+        >>> import torch.nn as nn
+        >>> class MyModule2(nn.Module):
+        >>>
+        >>>     # convert pred to fp32
+        >>>     @force_fp32(apply_to=('pred', ))
+        >>>     def post_process(self, pred, others):
+        >>>         pass
+    """
+
+    def force_fp32_wrapper(old_func):
+
+        @functools.wraps(old_func)
+        def new_func(*args, **kwargs):
+            # check if the module has set the attribute `fp16_enabled`, if not,
+            # just fallback to the original method.
+            if not isinstance(args[0], torch.nn.Module):
+                raise TypeError('@force_fp32 can only be used to decorate the '
+                                'method of nn.Module')
+            if not (hasattr(args[0], 'fp16_enabled') and args[0].fp16_enabled):
+                return old_func(*args, **kwargs)
+            # get the arg spec of the decorated method
+            args_info = getfullargspec(old_func)
+            # get the argument names to be casted
+            args_to_cast = args_info.args if apply_to is None else apply_to
+            # convert the args that need to be processed
+            new_args = []
+            if args:
+                arg_names = args_info.args[:len(args)]
+                for i, arg_name in enumerate(arg_names):
+                    if arg_name in args_to_cast:
+                        new_args.append(
+                            cast_tensor_type(args[i], torch.half, torch.float))
+                    else:
+                        new_args.append(args[i])
+            # convert the kwargs that need to be processed
+            new_kwargs = dict()
+            if kwargs:
+                for arg_name, arg_value in kwargs.items():
+                    if arg_name in args_to_cast:
+                        new_kwargs[arg_name] = cast_tensor_type(
+                            arg_value, torch.half, torch.float)
+                    else:
+                        new_kwargs[arg_name] = arg_value
+            # apply converted arguments to the decorated method
+            if (TORCH_VERSION != 'parrots' and
+                    digit_version(TORCH_VERSION) >= digit_version('1.6.0')):
+                with autocast(enabled=False):
+                    output = old_func(*new_args, **new_kwargs)
+            else:
+                output = old_func(*new_args, **new_kwargs)
+            # cast the results back to fp32 if necessary
+            if out_fp16:
+                output = cast_tensor_type(output, torch.float, torch.half)
+            return output
+
+        return new_func
+
+    return force_fp32_wrapper
+
+
+def allreduce_grads(params, coalesce=True, bucket_size_mb=-1):
+    warnings.warning(
+        '"mmcv.runner.fp16_utils.allreduce_grads" is deprecated, and will be '
+        'removed in v2.8. Please switch to "mmcv.runner.allreduce_grads')
+    _allreduce_grads(params, coalesce=coalesce, bucket_size_mb=bucket_size_mb)
+
+
+def wrap_fp16_model(model):
+    """Wrap the FP32 model to FP16.
+
+    If you are using PyTorch >= 1.6, torch.cuda.amp is used as the
+    backend, otherwise, original mmcv implementation will be adopted.
+
+    For PyTorch >= 1.6, this function will
+    1. Set fp16 flag inside the model to True.
+
+    Otherwise:
+    1. Convert FP32 model to FP16.
+    2. Remain some necessary layers to be FP32, e.g., normalization layers.
+    3. Set `fp16_enabled` flag inside the model to True.
+
+    Args:
+        model (nn.Module): Model in FP32.
+    """
+    if (TORCH_VERSION == 'parrots'
+            or digit_version(TORCH_VERSION) < digit_version('1.6.0')):
+        # convert model to fp16
+        model.half()
+        # patch the normalization layers to make it work in fp32 mode
+        patch_norm_fp32(model)
+    # set `fp16_enabled` flag
+    for m in model.modules():
+        if hasattr(m, 'fp16_enabled'):
+            m.fp16_enabled = True
+
+
+def patch_norm_fp32(module):
+    """Recursively convert normalization layers from FP16 to FP32.
+
+    Args:
+        module (nn.Module): The modules to be converted in FP16.
+
+    Returns:
+        nn.Module: The converted module, the normalization layers have been
+            converted to FP32.
+    """
+    if isinstance(module, (nn.modules.batchnorm._BatchNorm, nn.GroupNorm)):
+        module.float()
+        if isinstance(module, nn.GroupNorm) or torch.__version__ < '1.3':
+            module.forward = patch_forward_method(module.forward, torch.half,
+                                                  torch.float)
+    for child in module.children():
+        patch_norm_fp32(child)
+    return module
+
+
+def patch_forward_method(func, src_type, dst_type, convert_output=True):
+    """Patch the forward method of a module.
+
+    Args:
+        func (callable): The original forward method.
+        src_type (torch.dtype): Type of input arguments to be converted from.
+        dst_type (torch.dtype): Type of input arguments to be converted to.
+        convert_output (bool): Whether to convert the output back to src_type.
+
+    Returns:
+        callable: The patched forward method.
+    """
+
+    def new_forward(*args, **kwargs):
+        output = func(*cast_tensor_type(args, src_type, dst_type),
+                      **cast_tensor_type(kwargs, src_type, dst_type))
+        if convert_output:
+            output = cast_tensor_type(output, dst_type, src_type)
+        return output
+
+    return new_forward
+
+
+class LossScaler:
+    """Class that manages loss scaling in mixed precision training which
+    supports both dynamic or static mode.
+
+    The implementation refers to
+    https://github.com/NVIDIA/apex/blob/master/apex/fp16_utils/loss_scaler.py.
+    Indirectly, by supplying ``mode='dynamic'`` for dynamic loss scaling.
+    It's important to understand how :class:`LossScaler` operates.
+    Loss scaling is designed to combat the problem of underflowing
+    gradients encountered at long times when training fp16 networks.
+    Dynamic loss scaling begins by attempting a very high loss
+    scale.  Ironically, this may result in OVERflowing gradients.
+    If overflowing gradients are encountered, :class:`FP16_Optimizer` then
+    skips the update step for this particular iteration/minibatch,
+    and :class:`LossScaler` adjusts the loss scale to a lower value.
+    If a certain number of iterations occur without overflowing gradients
+    detected,:class:`LossScaler` increases the loss scale once more.
+    In this way :class:`LossScaler` attempts to "ride the edge" of always
+    using the highest loss scale possible without incurring overflow.
+
+    Args:
+        init_scale (float): Initial loss scale value, default: 2**32.
+        scale_factor (float): Factor used when adjusting the loss scale.
+            Default: 2.
+        mode (str): Loss scaling mode. 'dynamic' or 'static'
+        scale_window (int): Number of consecutive iterations without an
+            overflow to wait before increasing the loss scale. Default: 1000.
+    """
+
+    def __init__(self,
+                 init_scale=2**32,
+                 mode='dynamic',
+                 scale_factor=2.,
+                 scale_window=1000):
+        self.cur_scale = init_scale
+        self.cur_iter = 0
+        assert mode in ('dynamic',
+                        'static'), 'mode can only be dynamic or static'
+        self.mode = mode
+        self.last_overflow_iter = -1
+        self.scale_factor = scale_factor
+        self.scale_window = scale_window
+
+    def has_overflow(self, params):
+        """Check if params contain overflow."""
+        if self.mode != 'dynamic':
+            return False
+        for p in params:
+            if p.grad is not None and LossScaler._has_inf_or_nan(p.grad.data):
+                return True
+        return False
+
+    def _has_inf_or_nan(x):
+        """Check if params contain NaN."""
+        try:
+            cpu_sum = float(x.float().sum())
+        except RuntimeError as instance:
+            if 'value cannot be converted' not in instance.args[0]:
+                raise
+            return True
+        else:
+            if cpu_sum == float('inf') or cpu_sum == -float('inf') \
+                    or cpu_sum != cpu_sum:
+                return True
+            return False
+
+    def update_scale(self, overflow):
+        """update the current loss scale value when overflow happens."""
+        if self.mode != 'dynamic':
+            return
+        if overflow:
+            self.cur_scale = max(self.cur_scale / self.scale_factor, 1)
+            self.last_overflow_iter = self.cur_iter
+        else:
+            if (self.cur_iter - self.last_overflow_iter) % \
+                    self.scale_window == 0:
+                self.cur_scale *= self.scale_factor
+        self.cur_iter += 1
+
+    def state_dict(self):
+        """Returns the state of the scaler as a :class:`dict`."""
+        return dict(
+            cur_scale=self.cur_scale,
+            cur_iter=self.cur_iter,
+            mode=self.mode,
+            last_overflow_iter=self.last_overflow_iter,
+            scale_factor=self.scale_factor,
+            scale_window=self.scale_window)
+
+    def load_state_dict(self, state_dict):
+        """Loads the loss_scaler state dict.
+
+        Args:
+           state_dict (dict): scaler state.
+        """
+        self.cur_scale = state_dict['cur_scale']
+        self.cur_iter = state_dict['cur_iter']
+        self.mode = state_dict['mode']
+        self.last_overflow_iter = state_dict['last_overflow_iter']
+        self.scale_factor = state_dict['scale_factor']
+        self.scale_window = state_dict['scale_window']
+
+    @property
+    def loss_scale(self):
+        return self.cur_scale
diff --git a/annotator/uniformer/mmcv/runner/hooks/__init__.py b/annotator/uniformer/mmcv/runner/hooks/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..915af28cefab14a14c1188ed861161080fd138a3
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/__init__.py
@@ -0,0 +1,29 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .checkpoint import CheckpointHook
+from .closure import ClosureHook
+from .ema import EMAHook
+from .evaluation import DistEvalHook, EvalHook
+from .hook import HOOKS, Hook
+from .iter_timer import IterTimerHook
+from .logger import (DvcliveLoggerHook, LoggerHook, MlflowLoggerHook,
+                     NeptuneLoggerHook, PaviLoggerHook, TensorboardLoggerHook,
+                     TextLoggerHook, WandbLoggerHook)
+from .lr_updater import LrUpdaterHook
+from .memory import EmptyCacheHook
+from .momentum_updater import MomentumUpdaterHook
+from .optimizer import (Fp16OptimizerHook, GradientCumulativeFp16OptimizerHook,
+                        GradientCumulativeOptimizerHook, OptimizerHook)
+from .profiler import ProfilerHook
+from .sampler_seed import DistSamplerSeedHook
+from .sync_buffer import SyncBuffersHook
+
+__all__ = [
+    'HOOKS', 'Hook', 'CheckpointHook', 'ClosureHook', 'LrUpdaterHook',
+    'OptimizerHook', 'Fp16OptimizerHook', 'IterTimerHook',
+    'DistSamplerSeedHook', 'EmptyCacheHook', 'LoggerHook', 'MlflowLoggerHook',
+    'PaviLoggerHook', 'TextLoggerHook', 'TensorboardLoggerHook',
+    'NeptuneLoggerHook', 'WandbLoggerHook', 'DvcliveLoggerHook',
+    'MomentumUpdaterHook', 'SyncBuffersHook', 'EMAHook', 'EvalHook',
+    'DistEvalHook', 'ProfilerHook', 'GradientCumulativeOptimizerHook',
+    'GradientCumulativeFp16OptimizerHook'
+]
diff --git a/annotator/uniformer/mmcv/runner/hooks/checkpoint.py b/annotator/uniformer/mmcv/runner/hooks/checkpoint.py
new file mode 100644
index 0000000000000000000000000000000000000000..6af3fae43ac4b35532641a81eb13557edfc7dfba
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/checkpoint.py
@@ -0,0 +1,167 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import os.path as osp
+import warnings
+
+from annotator.uniformer.mmcv.fileio import FileClient
+from ..dist_utils import allreduce_params, master_only
+from .hook import HOOKS, Hook
+
+
+@HOOKS.register_module()
+class CheckpointHook(Hook):
+    """Save checkpoints periodically.
+
+    Args:
+        interval (int): The saving period. If ``by_epoch=True``, interval
+            indicates epochs, otherwise it indicates iterations.
+            Default: -1, which means "never".
+        by_epoch (bool): Saving checkpoints by epoch or by iteration.
+            Default: True.
+        save_optimizer (bool): Whether to save optimizer state_dict in the
+            checkpoint. It is usually used for resuming experiments.
+            Default: True.
+        out_dir (str, optional): The root directory to save checkpoints. If not
+            specified, ``runner.work_dir`` will be used by default. If
+            specified, the ``out_dir`` will be the concatenation of ``out_dir``
+            and the last level directory of ``runner.work_dir``.
+            `Changed in version 1.3.16.`
+        max_keep_ckpts (int, optional): The maximum checkpoints to keep.
+            In some cases we want only the latest few checkpoints and would
+            like to delete old ones to save the disk space.
+            Default: -1, which means unlimited.
+        save_last (bool, optional): Whether to force the last checkpoint to be
+            saved regardless of interval. Default: True.
+        sync_buffer (bool, optional): Whether to synchronize buffers in
+            different gpus. Default: False.
+        file_client_args (dict, optional): Arguments to instantiate a
+            FileClient. See :class:`mmcv.fileio.FileClient` for details.
+            Default: None.
+            `New in version 1.3.16.`
+
+    .. warning::
+        Before v1.3.16, the ``out_dir`` argument indicates the path where the
+        checkpoint is stored. However, since v1.3.16, ``out_dir`` indicates the
+        root directory and the final path to save checkpoint is the
+        concatenation of ``out_dir`` and the last level directory of
+        ``runner.work_dir``. Suppose the value of ``out_dir`` is "/path/of/A"
+        and the value of ``runner.work_dir`` is "/path/of/B", then the final
+        path will be "/path/of/A/B".
+    """
+
+    def __init__(self,
+                 interval=-1,
+                 by_epoch=True,
+                 save_optimizer=True,
+                 out_dir=None,
+                 max_keep_ckpts=-1,
+                 save_last=True,
+                 sync_buffer=False,
+                 file_client_args=None,
+                 **kwargs):
+        self.interval = interval
+        self.by_epoch = by_epoch
+        self.save_optimizer = save_optimizer
+        self.out_dir = out_dir
+        self.max_keep_ckpts = max_keep_ckpts
+        self.save_last = save_last
+        self.args = kwargs
+        self.sync_buffer = sync_buffer
+        self.file_client_args = file_client_args
+
+    def before_run(self, runner):
+        if not self.out_dir:
+            self.out_dir = runner.work_dir
+
+        self.file_client = FileClient.infer_client(self.file_client_args,
+                                                   self.out_dir)
+
+        # if `self.out_dir` is not equal to `runner.work_dir`, it means that
+        # `self.out_dir` is set so the final `self.out_dir` is the
+        # concatenation of `self.out_dir` and the last level directory of
+        # `runner.work_dir`
+        if self.out_dir != runner.work_dir:
+            basename = osp.basename(runner.work_dir.rstrip(osp.sep))
+            self.out_dir = self.file_client.join_path(self.out_dir, basename)
+
+        runner.logger.info((f'Checkpoints will be saved to {self.out_dir} by '
+                            f'{self.file_client.name}.'))
+
+        # disable the create_symlink option because some file backends do not
+        # allow to create a symlink
+        if 'create_symlink' in self.args:
+            if self.args[
+                    'create_symlink'] and not self.file_client.allow_symlink:
+                self.args['create_symlink'] = False
+                warnings.warn(
+                    ('create_symlink is set as True by the user but is changed'
+                     'to be False because creating symbolic link is not '
+                     f'allowed in {self.file_client.name}'))
+        else:
+            self.args['create_symlink'] = self.file_client.allow_symlink
+
+    def after_train_epoch(self, runner):
+        if not self.by_epoch:
+            return
+
+        # save checkpoint for following cases:
+        # 1. every ``self.interval`` epochs
+        # 2. reach the last epoch of training
+        if self.every_n_epochs(
+                runner, self.interval) or (self.save_last
+                                           and self.is_last_epoch(runner)):
+            runner.logger.info(
+                f'Saving checkpoint at {runner.epoch + 1} epochs')
+            if self.sync_buffer:
+                allreduce_params(runner.model.buffers())
+            self._save_checkpoint(runner)
+
+    @master_only
+    def _save_checkpoint(self, runner):
+        """Save the current checkpoint and delete unwanted checkpoint."""
+        runner.save_checkpoint(
+            self.out_dir, save_optimizer=self.save_optimizer, **self.args)
+        if runner.meta is not None:
+            if self.by_epoch:
+                cur_ckpt_filename = self.args.get(
+                    'filename_tmpl', 'epoch_{}.pth').format(runner.epoch + 1)
+            else:
+                cur_ckpt_filename = self.args.get(
+                    'filename_tmpl', 'iter_{}.pth').format(runner.iter + 1)
+            runner.meta.setdefault('hook_msgs', dict())
+            runner.meta['hook_msgs']['last_ckpt'] = self.file_client.join_path(
+                self.out_dir, cur_ckpt_filename)
+        # remove other checkpoints
+        if self.max_keep_ckpts > 0:
+            if self.by_epoch:
+                name = 'epoch_{}.pth'
+                current_ckpt = runner.epoch + 1
+            else:
+                name = 'iter_{}.pth'
+                current_ckpt = runner.iter + 1
+            redundant_ckpts = range(
+                current_ckpt - self.max_keep_ckpts * self.interval, 0,
+                -self.interval)
+            filename_tmpl = self.args.get('filename_tmpl', name)
+            for _step in redundant_ckpts:
+                ckpt_path = self.file_client.join_path(
+                    self.out_dir, filename_tmpl.format(_step))
+                if self.file_client.isfile(ckpt_path):
+                    self.file_client.remove(ckpt_path)
+                else:
+                    break
+
+    def after_train_iter(self, runner):
+        if self.by_epoch:
+            return
+
+        # save checkpoint for following cases:
+        # 1. every ``self.interval`` iterations
+        # 2. reach the last iteration of training
+        if self.every_n_iters(
+                runner, self.interval) or (self.save_last
+                                           and self.is_last_iter(runner)):
+            runner.logger.info(
+                f'Saving checkpoint at {runner.iter + 1} iterations')
+            if self.sync_buffer:
+                allreduce_params(runner.model.buffers())
+            self._save_checkpoint(runner)
diff --git a/annotator/uniformer/mmcv/runner/hooks/closure.py b/annotator/uniformer/mmcv/runner/hooks/closure.py
new file mode 100644
index 0000000000000000000000000000000000000000..b955f81f425be4ac3e6bb3f4aac653887989e872
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/closure.py
@@ -0,0 +1,11 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .hook import HOOKS, Hook
+
+
+@HOOKS.register_module()
+class ClosureHook(Hook):
+
+    def __init__(self, fn_name, fn):
+        assert hasattr(self, fn_name)
+        assert callable(fn)
+        setattr(self, fn_name, fn)
diff --git a/annotator/uniformer/mmcv/runner/hooks/ema.py b/annotator/uniformer/mmcv/runner/hooks/ema.py
new file mode 100644
index 0000000000000000000000000000000000000000..15c7e68088f019802a59e7ae41cc1fe0c7f28f96
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/ema.py
@@ -0,0 +1,89 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from ...parallel import is_module_wrapper
+from ..hooks.hook import HOOKS, Hook
+
+
+@HOOKS.register_module()
+class EMAHook(Hook):
+    r"""Exponential Moving Average Hook.
+
+    Use Exponential Moving Average on all parameters of model in training
+    process. All parameters have a ema backup, which update by the formula
+    as below. EMAHook takes priority over EvalHook and CheckpointSaverHook.
+
+        .. math::
+
+            \text{Xema\_{t+1}} = (1 - \text{momentum}) \times
+            \text{Xema\_{t}} +  \text{momentum} \times X_t
+
+    Args:
+        momentum (float): The momentum used for updating ema parameter.
+            Defaults to 0.0002.
+        interval (int): Update ema parameter every interval iteration.
+            Defaults to 1.
+        warm_up (int): During first warm_up steps, we may use smaller momentum
+            to update ema parameters more slowly. Defaults to 100.
+        resume_from (str): The checkpoint path. Defaults to None.
+    """
+
+    def __init__(self,
+                 momentum=0.0002,
+                 interval=1,
+                 warm_up=100,
+                 resume_from=None):
+        assert isinstance(interval, int) and interval > 0
+        self.warm_up = warm_up
+        self.interval = interval
+        assert momentum > 0 and momentum < 1
+        self.momentum = momentum**interval
+        self.checkpoint = resume_from
+
+    def before_run(self, runner):
+        """To resume model with it's ema parameters more friendly.
+
+        Register ema parameter as ``named_buffer`` to model
+        """
+        model = runner.model
+        if is_module_wrapper(model):
+            model = model.module
+        self.param_ema_buffer = {}
+        self.model_parameters = dict(model.named_parameters(recurse=True))
+        for name, value in self.model_parameters.items():
+            # "." is not allowed in module's buffer name
+            buffer_name = f"ema_{name.replace('.', '_')}"
+            self.param_ema_buffer[name] = buffer_name
+            model.register_buffer(buffer_name, value.data.clone())
+        self.model_buffers = dict(model.named_buffers(recurse=True))
+        if self.checkpoint is not None:
+            runner.resume(self.checkpoint)
+
+    def after_train_iter(self, runner):
+        """Update ema parameter every self.interval iterations."""
+        curr_step = runner.iter
+        # We warm up the momentum considering the instability at beginning
+        momentum = min(self.momentum,
+                       (1 + curr_step) / (self.warm_up + curr_step))
+        if curr_step % self.interval != 0:
+            return
+        for name, parameter in self.model_parameters.items():
+            buffer_name = self.param_ema_buffer[name]
+            buffer_parameter = self.model_buffers[buffer_name]
+            buffer_parameter.mul_(1 - momentum).add_(momentum, parameter.data)
+
+    def after_train_epoch(self, runner):
+        """We load parameter values from ema backup to model before the
+        EvalHook."""
+        self._swap_ema_parameters()
+
+    def before_train_epoch(self, runner):
+        """We recover model's parameter from ema backup after last epoch's
+        EvalHook."""
+        self._swap_ema_parameters()
+
+    def _swap_ema_parameters(self):
+        """Swap the parameter of model with parameter in ema_buffer."""
+        for name, value in self.model_parameters.items():
+            temp = value.data.clone()
+            ema_buffer = self.model_buffers[self.param_ema_buffer[name]]
+            value.data.copy_(ema_buffer.data)
+            ema_buffer.data.copy_(temp)
diff --git a/annotator/uniformer/mmcv/runner/hooks/evaluation.py b/annotator/uniformer/mmcv/runner/hooks/evaluation.py
new file mode 100644
index 0000000000000000000000000000000000000000..4d00999ce5665c53bded8de9e084943eee2d230d
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/evaluation.py
@@ -0,0 +1,509 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import os.path as osp
+import warnings
+from math import inf
+
+import torch.distributed as dist
+from torch.nn.modules.batchnorm import _BatchNorm
+from torch.utils.data import DataLoader
+
+from annotator.uniformer.mmcv.fileio import FileClient
+from annotator.uniformer.mmcv.utils import is_seq_of
+from .hook import Hook
+from .logger import LoggerHook
+
+
+class EvalHook(Hook):
+    """Non-Distributed evaluation hook.
+
+    This hook will regularly perform evaluation in a given interval when
+    performing in non-distributed environment.
+
+    Args:
+        dataloader (DataLoader): A PyTorch dataloader, whose dataset has
+            implemented ``evaluate`` function.
+        start (int | None, optional): Evaluation starting epoch. It enables
+            evaluation before the training starts if ``start`` <= the resuming
+            epoch. If None, whether to evaluate is merely decided by
+            ``interval``. Default: None.
+        interval (int): Evaluation interval. Default: 1.
+        by_epoch (bool): Determine perform evaluation by epoch or by iteration.
+            If set to True, it will perform by epoch. Otherwise, by iteration.
+            Default: True.
+        save_best (str, optional): If a metric is specified, it would measure
+            the best checkpoint during evaluation. The information about best
+            checkpoint would be saved in ``runner.meta['hook_msgs']`` to keep
+            best score value and best checkpoint path, which will be also
+            loaded when resume checkpoint. Options are the evaluation metrics
+            on the test dataset. e.g., ``bbox_mAP``, ``segm_mAP`` for bbox
+            detection and instance segmentation. ``AR@100`` for proposal
+            recall. If ``save_best`` is ``auto``, the first key of the returned
+            ``OrderedDict`` result will be used. Default: None.
+        rule (str | None, optional): Comparison rule for best score. If set to
+            None, it will infer a reasonable rule. Keys such as 'acc', 'top'
+            .etc will be inferred by 'greater' rule. Keys contain 'loss' will
+            be inferred by 'less' rule. Options are 'greater', 'less', None.
+            Default: None.
+        test_fn (callable, optional): test a model with samples from a
+            dataloader, and return the test results. If ``None``, the default
+            test function ``mmcv.engine.single_gpu_test`` will be used.
+            (default: ``None``)
+        greater_keys (List[str] | None, optional): Metric keys that will be
+            inferred by 'greater' comparison rule. If ``None``,
+            _default_greater_keys will be used. (default: ``None``)
+        less_keys (List[str] | None, optional): Metric keys that will be
+            inferred by 'less' comparison rule. If ``None``, _default_less_keys
+            will be used. (default: ``None``)
+        out_dir (str, optional): The root directory to save checkpoints. If not
+            specified, `runner.work_dir` will be used by default. If specified,
+            the `out_dir` will be the concatenation of `out_dir` and the last
+            level directory of `runner.work_dir`.
+            `New in version 1.3.16.`
+        file_client_args (dict): Arguments to instantiate a FileClient.
+            See :class:`mmcv.fileio.FileClient` for details. Default: None.
+            `New in version 1.3.16.`
+        **eval_kwargs: Evaluation arguments fed into the evaluate function of
+            the dataset.
+
+    Notes:
+        If new arguments are added for EvalHook, tools/test.py,
+        tools/eval_metric.py may be affected.
+    """
+
+    # Since the key for determine greater or less is related to the downstream
+    # tasks, downstream repos may need to overwrite the following inner
+    # variable accordingly.
+
+    rule_map = {'greater': lambda x, y: x > y, 'less': lambda x, y: x < y}
+    init_value_map = {'greater': -inf, 'less': inf}
+    _default_greater_keys = [
+        'acc', 'top', 'AR@', 'auc', 'precision', 'mAP', 'mDice', 'mIoU',
+        'mAcc', 'aAcc'
+    ]
+    _default_less_keys = ['loss']
+
+    def __init__(self,
+                 dataloader,
+                 start=None,
+                 interval=1,
+                 by_epoch=True,
+                 save_best=None,
+                 rule=None,
+                 test_fn=None,
+                 greater_keys=None,
+                 less_keys=None,
+                 out_dir=None,
+                 file_client_args=None,
+                 **eval_kwargs):
+        if not isinstance(dataloader, DataLoader):
+            raise TypeError(f'dataloader must be a pytorch DataLoader, '
+                            f'but got {type(dataloader)}')
+
+        if interval <= 0:
+            raise ValueError(f'interval must be a positive number, '
+                             f'but got {interval}')
+
+        assert isinstance(by_epoch, bool), '``by_epoch`` should be a boolean'
+
+        if start is not None and start < 0:
+            raise ValueError(f'The evaluation start epoch {start} is smaller '
+                             f'than 0')
+
+        self.dataloader = dataloader
+        self.interval = interval
+        self.start = start
+        self.by_epoch = by_epoch
+
+        assert isinstance(save_best, str) or save_best is None, \
+            '""save_best"" should be a str or None ' \
+            f'rather than {type(save_best)}'
+        self.save_best = save_best
+        self.eval_kwargs = eval_kwargs
+        self.initial_flag = True
+
+        if test_fn is None:
+            from annotator.uniformer.mmcv.engine import single_gpu_test
+            self.test_fn = single_gpu_test
+        else:
+            self.test_fn = test_fn
+
+        if greater_keys is None:
+            self.greater_keys = self._default_greater_keys
+        else:
+            if not isinstance(greater_keys, (list, tuple)):
+                greater_keys = (greater_keys, )
+            assert is_seq_of(greater_keys, str)
+            self.greater_keys = greater_keys
+
+        if less_keys is None:
+            self.less_keys = self._default_less_keys
+        else:
+            if not isinstance(less_keys, (list, tuple)):
+                less_keys = (less_keys, )
+            assert is_seq_of(less_keys, str)
+            self.less_keys = less_keys
+
+        if self.save_best is not None:
+            self.best_ckpt_path = None
+            self._init_rule(rule, self.save_best)
+
+        self.out_dir = out_dir
+        self.file_client_args = file_client_args
+
+    def _init_rule(self, rule, key_indicator):
+        """Initialize rule, key_indicator, comparison_func, and best score.
+
+        Here is the rule to determine which rule is used for key indicator
+        when the rule is not specific (note that the key indicator matching
+        is case-insensitive):
+        1. If the key indicator is in ``self.greater_keys``, the rule will be
+           specified as 'greater'.
+        2. Or if the key indicator is in ``self.less_keys``, the rule will be
+           specified as 'less'.
+        3. Or if the key indicator is equal to the substring in any one item
+           in ``self.greater_keys``, the rule will be specified as 'greater'.
+        4. Or if the key indicator is equal to the substring in any one item
+           in ``self.less_keys``, the rule will be specified as 'less'.
+
+        Args:
+            rule (str | None): Comparison rule for best score.
+            key_indicator (str | None): Key indicator to determine the
+                comparison rule.
+        """
+        if rule not in self.rule_map and rule is not None:
+            raise KeyError(f'rule must be greater, less or None, '
+                           f'but got {rule}.')
+
+        if rule is None:
+            if key_indicator != 'auto':
+                # `_lc` here means we use the lower case of keys for
+                # case-insensitive matching
+                key_indicator_lc = key_indicator.lower()
+                greater_keys = [key.lower() for key in self.greater_keys]
+                less_keys = [key.lower() for key in self.less_keys]
+
+                if key_indicator_lc in greater_keys:
+                    rule = 'greater'
+                elif key_indicator_lc in less_keys:
+                    rule = 'less'
+                elif any(key in key_indicator_lc for key in greater_keys):
+                    rule = 'greater'
+                elif any(key in key_indicator_lc for key in less_keys):
+                    rule = 'less'
+                else:
+                    raise ValueError(f'Cannot infer the rule for key '
+                                     f'{key_indicator}, thus a specific rule '
+                                     f'must be specified.')
+        self.rule = rule
+        self.key_indicator = key_indicator
+        if self.rule is not None:
+            self.compare_func = self.rule_map[self.rule]
+
+    def before_run(self, runner):
+        if not self.out_dir:
+            self.out_dir = runner.work_dir
+
+        self.file_client = FileClient.infer_client(self.file_client_args,
+                                                   self.out_dir)
+
+        # if `self.out_dir` is not equal to `runner.work_dir`, it means that
+        # `self.out_dir` is set so the final `self.out_dir` is the
+        # concatenation of `self.out_dir` and the last level directory of
+        # `runner.work_dir`
+        if self.out_dir != runner.work_dir:
+            basename = osp.basename(runner.work_dir.rstrip(osp.sep))
+            self.out_dir = self.file_client.join_path(self.out_dir, basename)
+            runner.logger.info(
+                (f'The best checkpoint will be saved to {self.out_dir} by '
+                 f'{self.file_client.name}'))
+
+        if self.save_best is not None:
+            if runner.meta is None:
+                warnings.warn('runner.meta is None. Creating an empty one.')
+                runner.meta = dict()
+            runner.meta.setdefault('hook_msgs', dict())
+            self.best_ckpt_path = runner.meta['hook_msgs'].get(
+                'best_ckpt', None)
+
+    def before_train_iter(self, runner):
+        """Evaluate the model only at the start of training by iteration."""
+        if self.by_epoch or not self.initial_flag:
+            return
+        if self.start is not None and runner.iter >= self.start:
+            self.after_train_iter(runner)
+        self.initial_flag = False
+
+    def before_train_epoch(self, runner):
+        """Evaluate the model only at the start of training by epoch."""
+        if not (self.by_epoch and self.initial_flag):
+            return
+        if self.start is not None and runner.epoch >= self.start:
+            self.after_train_epoch(runner)
+        self.initial_flag = False
+
+    def after_train_iter(self, runner):
+        """Called after every training iter to evaluate the results."""
+        if not self.by_epoch and self._should_evaluate(runner):
+            # Because the priority of EvalHook is higher than LoggerHook, the
+            # training log and the evaluating log are mixed. Therefore,
+            # we need to dump the training log and clear it before evaluating
+            # log is generated. In addition, this problem will only appear in
+            # `IterBasedRunner` whose `self.by_epoch` is False, because
+            # `EpochBasedRunner` whose `self.by_epoch` is True calls
+            # `_do_evaluate` in `after_train_epoch` stage, and at this stage
+            # the training log has been printed, so it will not cause any
+            # problem. more details at
+            # https://github.com/open-mmlab/mmsegmentation/issues/694
+            for hook in runner._hooks:
+                if isinstance(hook, LoggerHook):
+                    hook.after_train_iter(runner)
+            runner.log_buffer.clear()
+
+            self._do_evaluate(runner)
+
+    def after_train_epoch(self, runner):
+        """Called after every training epoch to evaluate the results."""
+        if self.by_epoch and self._should_evaluate(runner):
+            self._do_evaluate(runner)
+
+    def _do_evaluate(self, runner):
+        """perform evaluation and save ckpt."""
+        results = self.test_fn(runner.model, self.dataloader)
+        runner.log_buffer.output['eval_iter_num'] = len(self.dataloader)
+        key_score = self.evaluate(runner, results)
+        # the key_score may be `None` so it needs to skip the action to save
+        # the best checkpoint
+        if self.save_best and key_score:
+            self._save_ckpt(runner, key_score)
+
+    def _should_evaluate(self, runner):
+        """Judge whether to perform evaluation.
+
+        Here is the rule to judge whether to perform evaluation:
+        1. It will not perform evaluation during the epoch/iteration interval,
+           which is determined by ``self.interval``.
+        2. It will not perform evaluation if the start time is larger than
+           current time.
+        3. It will not perform evaluation when current time is larger than
+           the start time but during epoch/iteration interval.
+
+        Returns:
+            bool: The flag indicating whether to perform evaluation.
+        """
+        if self.by_epoch:
+            current = runner.epoch
+            check_time = self.every_n_epochs
+        else:
+            current = runner.iter
+            check_time = self.every_n_iters
+
+        if self.start is None:
+            if not check_time(runner, self.interval):
+                # No evaluation during the interval.
+                return False
+        elif (current + 1) < self.start:
+            # No evaluation if start is larger than the current time.
+            return False
+        else:
+            # Evaluation only at epochs/iters 3, 5, 7...
+            # if start==3 and interval==2
+            if (current + 1 - self.start) % self.interval:
+                return False
+        return True
+
+    def _save_ckpt(self, runner, key_score):
+        """Save the best checkpoint.
+
+        It will compare the score according to the compare function, write
+        related information (best score, best checkpoint path) and save the
+        best checkpoint into ``work_dir``.
+        """
+        if self.by_epoch:
+            current = f'epoch_{runner.epoch + 1}'
+            cur_type, cur_time = 'epoch', runner.epoch + 1
+        else:
+            current = f'iter_{runner.iter + 1}'
+            cur_type, cur_time = 'iter', runner.iter + 1
+
+        best_score = runner.meta['hook_msgs'].get(
+            'best_score', self.init_value_map[self.rule])
+        if self.compare_func(key_score, best_score):
+            best_score = key_score
+            runner.meta['hook_msgs']['best_score'] = best_score
+
+            if self.best_ckpt_path and self.file_client.isfile(
+                    self.best_ckpt_path):
+                self.file_client.remove(self.best_ckpt_path)
+                runner.logger.info(
+                    (f'The previous best checkpoint {self.best_ckpt_path} was '
+                     'removed'))
+
+            best_ckpt_name = f'best_{self.key_indicator}_{current}.pth'
+            self.best_ckpt_path = self.file_client.join_path(
+                self.out_dir, best_ckpt_name)
+            runner.meta['hook_msgs']['best_ckpt'] = self.best_ckpt_path
+
+            runner.save_checkpoint(
+                self.out_dir, best_ckpt_name, create_symlink=False)
+            runner.logger.info(
+                f'Now best checkpoint is saved as {best_ckpt_name}.')
+            runner.logger.info(
+                f'Best {self.key_indicator} is {best_score:0.4f} '
+                f'at {cur_time} {cur_type}.')
+
+    def evaluate(self, runner, results):
+        """Evaluate the results.
+
+        Args:
+            runner (:obj:`mmcv.Runner`): The underlined training runner.
+            results (list): Output results.
+        """
+        eval_res = self.dataloader.dataset.evaluate(
+            results, logger=runner.logger, **self.eval_kwargs)
+
+        for name, val in eval_res.items():
+            runner.log_buffer.output[name] = val
+        runner.log_buffer.ready = True
+
+        if self.save_best is not None:
+            # If the performance of model is pool, the `eval_res` may be an
+            # empty dict and it will raise exception when `self.save_best` is
+            # not None. More details at
+            # https://github.com/open-mmlab/mmdetection/issues/6265.
+            if not eval_res:
+                warnings.warn(
+                    'Since `eval_res` is an empty dict, the behavior to save '
+                    'the best checkpoint will be skipped in this evaluation.')
+                return None
+
+            if self.key_indicator == 'auto':
+                # infer from eval_results
+                self._init_rule(self.rule, list(eval_res.keys())[0])
+            return eval_res[self.key_indicator]
+
+        return None
+
+
+class DistEvalHook(EvalHook):
+    """Distributed evaluation hook.
+
+    This hook will regularly perform evaluation in a given interval when
+    performing in distributed environment.
+
+    Args:
+        dataloader (DataLoader): A PyTorch dataloader, whose dataset has
+            implemented ``evaluate`` function.
+        start (int | None, optional): Evaluation starting epoch. It enables
+            evaluation before the training starts if ``start`` <= the resuming
+            epoch. If None, whether to evaluate is merely decided by
+            ``interval``. Default: None.
+        interval (int): Evaluation interval. Default: 1.
+        by_epoch (bool): Determine perform evaluation by epoch or by iteration.
+            If set to True, it will perform by epoch. Otherwise, by iteration.
+            default: True.
+        save_best (str, optional): If a metric is specified, it would measure
+            the best checkpoint during evaluation. The information about best
+            checkpoint would be saved in ``runner.meta['hook_msgs']`` to keep
+            best score value and best checkpoint path, which will be also
+            loaded when resume checkpoint. Options are the evaluation metrics
+            on the test dataset. e.g., ``bbox_mAP``, ``segm_mAP`` for bbox
+            detection and instance segmentation. ``AR@100`` for proposal
+            recall. If ``save_best`` is ``auto``, the first key of the returned
+            ``OrderedDict`` result will be used. Default: None.
+        rule (str | None, optional): Comparison rule for best score. If set to
+            None, it will infer a reasonable rule. Keys such as 'acc', 'top'
+            .etc will be inferred by 'greater' rule. Keys contain 'loss' will
+            be inferred by 'less' rule. Options are 'greater', 'less', None.
+            Default: None.
+        test_fn (callable, optional): test a model with samples from a
+            dataloader in a multi-gpu manner, and return the test results. If
+            ``None``, the default test function ``mmcv.engine.multi_gpu_test``
+            will be used. (default: ``None``)
+        tmpdir (str | None): Temporary directory to save the results of all
+            processes. Default: None.
+        gpu_collect (bool): Whether to use gpu or cpu to collect results.
+            Default: False.
+        broadcast_bn_buffer (bool): Whether to broadcast the
+            buffer(running_mean and running_var) of rank 0 to other rank
+            before evaluation. Default: True.
+        out_dir (str, optional): The root directory to save checkpoints. If not
+            specified, `runner.work_dir` will be used by default. If specified,
+            the `out_dir` will be the concatenation of `out_dir` and the last
+            level directory of `runner.work_dir`.
+        file_client_args (dict): Arguments to instantiate a FileClient.
+            See :class:`mmcv.fileio.FileClient` for details. Default: None.
+        **eval_kwargs: Evaluation arguments fed into the evaluate function of
+            the dataset.
+    """
+
+    def __init__(self,
+                 dataloader,
+                 start=None,
+                 interval=1,
+                 by_epoch=True,
+                 save_best=None,
+                 rule=None,
+                 test_fn=None,
+                 greater_keys=None,
+                 less_keys=None,
+                 broadcast_bn_buffer=True,
+                 tmpdir=None,
+                 gpu_collect=False,
+                 out_dir=None,
+                 file_client_args=None,
+                 **eval_kwargs):
+
+        if test_fn is None:
+            from annotator.uniformer.mmcv.engine import multi_gpu_test
+            test_fn = multi_gpu_test
+
+        super().__init__(
+            dataloader,
+            start=start,
+            interval=interval,
+            by_epoch=by_epoch,
+            save_best=save_best,
+            rule=rule,
+            test_fn=test_fn,
+            greater_keys=greater_keys,
+            less_keys=less_keys,
+            out_dir=out_dir,
+            file_client_args=file_client_args,
+            **eval_kwargs)
+
+        self.broadcast_bn_buffer = broadcast_bn_buffer
+        self.tmpdir = tmpdir
+        self.gpu_collect = gpu_collect
+
+    def _do_evaluate(self, runner):
+        """perform evaluation and save ckpt."""
+        # Synchronization of BatchNorm's buffer (running_mean
+        # and running_var) is not supported in the DDP of pytorch,
+        # which may cause the inconsistent performance of models in
+        # different ranks, so we broadcast BatchNorm's buffers
+        # of rank 0 to other ranks to avoid this.
+        if self.broadcast_bn_buffer:
+            model = runner.model
+            for name, module in model.named_modules():
+                if isinstance(module,
+                              _BatchNorm) and module.track_running_stats:
+                    dist.broadcast(module.running_var, 0)
+                    dist.broadcast(module.running_mean, 0)
+
+        tmpdir = self.tmpdir
+        if tmpdir is None:
+            tmpdir = osp.join(runner.work_dir, '.eval_hook')
+
+        results = self.test_fn(
+            runner.model,
+            self.dataloader,
+            tmpdir=tmpdir,
+            gpu_collect=self.gpu_collect)
+        if runner.rank == 0:
+            print('\n')
+            runner.log_buffer.output['eval_iter_num'] = len(self.dataloader)
+            key_score = self.evaluate(runner, results)
+            # the key_score may be `None` so it needs to skip the action to
+            # save the best checkpoint
+            if self.save_best and key_score:
+                self._save_ckpt(runner, key_score)
diff --git a/annotator/uniformer/mmcv/runner/hooks/hook.py b/annotator/uniformer/mmcv/runner/hooks/hook.py
new file mode 100644
index 0000000000000000000000000000000000000000..b8855c107727ecf85b917c890fc8b7f6359238a4
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/hook.py
@@ -0,0 +1,92 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from annotator.uniformer.mmcv.utils import Registry, is_method_overridden
+
+HOOKS = Registry('hook')
+
+
+class Hook:
+    stages = ('before_run', 'before_train_epoch', 'before_train_iter',
+              'after_train_iter', 'after_train_epoch', 'before_val_epoch',
+              'before_val_iter', 'after_val_iter', 'after_val_epoch',
+              'after_run')
+
+    def before_run(self, runner):
+        pass
+
+    def after_run(self, runner):
+        pass
+
+    def before_epoch(self, runner):
+        pass
+
+    def after_epoch(self, runner):
+        pass
+
+    def before_iter(self, runner):
+        pass
+
+    def after_iter(self, runner):
+        pass
+
+    def before_train_epoch(self, runner):
+        self.before_epoch(runner)
+
+    def before_val_epoch(self, runner):
+        self.before_epoch(runner)
+
+    def after_train_epoch(self, runner):
+        self.after_epoch(runner)
+
+    def after_val_epoch(self, runner):
+        self.after_epoch(runner)
+
+    def before_train_iter(self, runner):
+        self.before_iter(runner)
+
+    def before_val_iter(self, runner):
+        self.before_iter(runner)
+
+    def after_train_iter(self, runner):
+        self.after_iter(runner)
+
+    def after_val_iter(self, runner):
+        self.after_iter(runner)
+
+    def every_n_epochs(self, runner, n):
+        return (runner.epoch + 1) % n == 0 if n > 0 else False
+
+    def every_n_inner_iters(self, runner, n):
+        return (runner.inner_iter + 1) % n == 0 if n > 0 else False
+
+    def every_n_iters(self, runner, n):
+        return (runner.iter + 1) % n == 0 if n > 0 else False
+
+    def end_of_epoch(self, runner):
+        return runner.inner_iter + 1 == len(runner.data_loader)
+
+    def is_last_epoch(self, runner):
+        return runner.epoch + 1 == runner._max_epochs
+
+    def is_last_iter(self, runner):
+        return runner.iter + 1 == runner._max_iters
+
+    def get_triggered_stages(self):
+        trigger_stages = set()
+        for stage in Hook.stages:
+            if is_method_overridden(stage, Hook, self):
+                trigger_stages.add(stage)
+
+        # some methods will be triggered in multi stages
+        # use this dict to map method to stages.
+        method_stages_map = {
+            'before_epoch': ['before_train_epoch', 'before_val_epoch'],
+            'after_epoch': ['after_train_epoch', 'after_val_epoch'],
+            'before_iter': ['before_train_iter', 'before_val_iter'],
+            'after_iter': ['after_train_iter', 'after_val_iter'],
+        }
+
+        for method, map_stages in method_stages_map.items():
+            if is_method_overridden(method, Hook, self):
+                trigger_stages.update(map_stages)
+
+        return [stage for stage in Hook.stages if stage in trigger_stages]
diff --git a/annotator/uniformer/mmcv/runner/hooks/iter_timer.py b/annotator/uniformer/mmcv/runner/hooks/iter_timer.py
new file mode 100644
index 0000000000000000000000000000000000000000..cfd5002fe85ffc6992155ac01003878064a1d9be
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/iter_timer.py
@@ -0,0 +1,18 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import time
+
+from .hook import HOOKS, Hook
+
+
+@HOOKS.register_module()
+class IterTimerHook(Hook):
+
+    def before_epoch(self, runner):
+        self.t = time.time()
+
+    def before_iter(self, runner):
+        runner.log_buffer.update({'data_time': time.time() - self.t})
+
+    def after_iter(self, runner):
+        runner.log_buffer.update({'time': time.time() - self.t})
+        self.t = time.time()
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/__init__.py b/annotator/uniformer/mmcv/runner/hooks/logger/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..a0b6b345640a895368ac8a647afef6f24333d90e
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/__init__.py
@@ -0,0 +1,15 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .base import LoggerHook
+from .dvclive import DvcliveLoggerHook
+from .mlflow import MlflowLoggerHook
+from .neptune import NeptuneLoggerHook
+from .pavi import PaviLoggerHook
+from .tensorboard import TensorboardLoggerHook
+from .text import TextLoggerHook
+from .wandb import WandbLoggerHook
+
+__all__ = [
+    'LoggerHook', 'MlflowLoggerHook', 'PaviLoggerHook',
+    'TensorboardLoggerHook', 'TextLoggerHook', 'WandbLoggerHook',
+    'NeptuneLoggerHook', 'DvcliveLoggerHook'
+]
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/base.py b/annotator/uniformer/mmcv/runner/hooks/logger/base.py
new file mode 100644
index 0000000000000000000000000000000000000000..f845256729458ced821762a1b8ef881e17ff9955
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/base.py
@@ -0,0 +1,166 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import numbers
+from abc import ABCMeta, abstractmethod
+
+import numpy as np
+import torch
+
+from ..hook import Hook
+
+
+class LoggerHook(Hook):
+    """Base class for logger hooks.
+
+    Args:
+        interval (int): Logging interval (every k iterations).
+        ignore_last (bool): Ignore the log of last iterations in each epoch
+            if less than `interval`.
+        reset_flag (bool): Whether to clear the output buffer after logging.
+        by_epoch (bool): Whether EpochBasedRunner is used.
+    """
+
+    __metaclass__ = ABCMeta
+
+    def __init__(self,
+                 interval=10,
+                 ignore_last=True,
+                 reset_flag=False,
+                 by_epoch=True):
+        self.interval = interval
+        self.ignore_last = ignore_last
+        self.reset_flag = reset_flag
+        self.by_epoch = by_epoch
+
+    @abstractmethod
+    def log(self, runner):
+        pass
+
+    @staticmethod
+    def is_scalar(val, include_np=True, include_torch=True):
+        """Tell the input variable is a scalar or not.
+
+        Args:
+            val: Input variable.
+            include_np (bool): Whether include 0-d np.ndarray as a scalar.
+            include_torch (bool): Whether include 0-d torch.Tensor as a scalar.
+
+        Returns:
+            bool: True or False.
+        """
+        if isinstance(val, numbers.Number):
+            return True
+        elif include_np and isinstance(val, np.ndarray) and val.ndim == 0:
+            return True
+        elif include_torch and isinstance(val, torch.Tensor) and len(val) == 1:
+            return True
+        else:
+            return False
+
+    def get_mode(self, runner):
+        if runner.mode == 'train':
+            if 'time' in runner.log_buffer.output:
+                mode = 'train'
+            else:
+                mode = 'val'
+        elif runner.mode == 'val':
+            mode = 'val'
+        else:
+            raise ValueError(f"runner mode should be 'train' or 'val', "
+                             f'but got {runner.mode}')
+        return mode
+
+    def get_epoch(self, runner):
+        if runner.mode == 'train':
+            epoch = runner.epoch + 1
+        elif runner.mode == 'val':
+            # normal val mode
+            # runner.epoch += 1 has been done before val workflow
+            epoch = runner.epoch
+        else:
+            raise ValueError(f"runner mode should be 'train' or 'val', "
+                             f'but got {runner.mode}')
+        return epoch
+
+    def get_iter(self, runner, inner_iter=False):
+        """Get the current training iteration step."""
+        if self.by_epoch and inner_iter:
+            current_iter = runner.inner_iter + 1
+        else:
+            current_iter = runner.iter + 1
+        return current_iter
+
+    def get_lr_tags(self, runner):
+        tags = {}
+        lrs = runner.current_lr()
+        if isinstance(lrs, dict):
+            for name, value in lrs.items():
+                tags[f'learning_rate/{name}'] = value[0]
+        else:
+            tags['learning_rate'] = lrs[0]
+        return tags
+
+    def get_momentum_tags(self, runner):
+        tags = {}
+        momentums = runner.current_momentum()
+        if isinstance(momentums, dict):
+            for name, value in momentums.items():
+                tags[f'momentum/{name}'] = value[0]
+        else:
+            tags['momentum'] = momentums[0]
+        return tags
+
+    def get_loggable_tags(self,
+                          runner,
+                          allow_scalar=True,
+                          allow_text=False,
+                          add_mode=True,
+                          tags_to_skip=('time', 'data_time')):
+        tags = {}
+        for var, val in runner.log_buffer.output.items():
+            if var in tags_to_skip:
+                continue
+            if self.is_scalar(val) and not allow_scalar:
+                continue
+            if isinstance(val, str) and not allow_text:
+                continue
+            if add_mode:
+                var = f'{self.get_mode(runner)}/{var}'
+            tags[var] = val
+        tags.update(self.get_lr_tags(runner))
+        tags.update(self.get_momentum_tags(runner))
+        return tags
+
+    def before_run(self, runner):
+        for hook in runner.hooks[::-1]:
+            if isinstance(hook, LoggerHook):
+                hook.reset_flag = True
+                break
+
+    def before_epoch(self, runner):
+        runner.log_buffer.clear()  # clear logs of last epoch
+
+    def after_train_iter(self, runner):
+        if self.by_epoch and self.every_n_inner_iters(runner, self.interval):
+            runner.log_buffer.average(self.interval)
+        elif not self.by_epoch and self.every_n_iters(runner, self.interval):
+            runner.log_buffer.average(self.interval)
+        elif self.end_of_epoch(runner) and not self.ignore_last:
+            # not precise but more stable
+            runner.log_buffer.average(self.interval)
+
+        if runner.log_buffer.ready:
+            self.log(runner)
+            if self.reset_flag:
+                runner.log_buffer.clear_output()
+
+    def after_train_epoch(self, runner):
+        if runner.log_buffer.ready:
+            self.log(runner)
+            if self.reset_flag:
+                runner.log_buffer.clear_output()
+
+    def after_val_epoch(self, runner):
+        runner.log_buffer.average()
+        self.log(runner)
+        if self.reset_flag:
+            runner.log_buffer.clear_output()
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/dvclive.py b/annotator/uniformer/mmcv/runner/hooks/logger/dvclive.py
new file mode 100644
index 0000000000000000000000000000000000000000..687cdc58c0336c92b1e4f9a410ba67ebaab2bc7a
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/dvclive.py
@@ -0,0 +1,58 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from ...dist_utils import master_only
+from ..hook import HOOKS
+from .base import LoggerHook
+
+
+@HOOKS.register_module()
+class DvcliveLoggerHook(LoggerHook):
+    """Class to log metrics with dvclive.
+
+    It requires `dvclive`_ to be installed.
+
+    Args:
+        path (str): Directory where dvclive will write TSV log files.
+        interval (int): Logging interval (every k iterations).
+            Default 10.
+        ignore_last (bool): Ignore the log of last iterations in each epoch
+            if less than `interval`.
+            Default: True.
+        reset_flag (bool): Whether to clear the output buffer after logging.
+            Default: True.
+        by_epoch (bool): Whether EpochBasedRunner is used.
+            Default: True.
+
+    .. _dvclive:
+        https://dvc.org/doc/dvclive
+    """
+
+    def __init__(self,
+                 path,
+                 interval=10,
+                 ignore_last=True,
+                 reset_flag=True,
+                 by_epoch=True):
+
+        super(DvcliveLoggerHook, self).__init__(interval, ignore_last,
+                                                reset_flag, by_epoch)
+        self.path = path
+        self.import_dvclive()
+
+    def import_dvclive(self):
+        try:
+            import dvclive
+        except ImportError:
+            raise ImportError(
+                'Please run "pip install dvclive" to install dvclive')
+        self.dvclive = dvclive
+
+    @master_only
+    def before_run(self, runner):
+        self.dvclive.init(self.path)
+
+    @master_only
+    def log(self, runner):
+        tags = self.get_loggable_tags(runner)
+        if tags:
+            for k, v in tags.items():
+                self.dvclive.log(k, v, step=self.get_iter(runner))
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/mlflow.py b/annotator/uniformer/mmcv/runner/hooks/logger/mlflow.py
new file mode 100644
index 0000000000000000000000000000000000000000..f9a72592be47b534ce22573775fd5a7e8e86d72d
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/mlflow.py
@@ -0,0 +1,78 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from ...dist_utils import master_only
+from ..hook import HOOKS
+from .base import LoggerHook
+
+
+@HOOKS.register_module()
+class MlflowLoggerHook(LoggerHook):
+
+    def __init__(self,
+                 exp_name=None,
+                 tags=None,
+                 log_model=True,
+                 interval=10,
+                 ignore_last=True,
+                 reset_flag=False,
+                 by_epoch=True):
+        """Class to log metrics and (optionally) a trained model to MLflow.
+
+        It requires `MLflow`_ to be installed.
+
+        Args:
+            exp_name (str, optional): Name of the experiment to be used.
+                Default None.
+                If not None, set the active experiment.
+                If experiment does not exist, an experiment with provided name
+                will be created.
+            tags (dict of str: str, optional): Tags for the current run.
+                Default None.
+                If not None, set tags for the current run.
+            log_model (bool, optional): Whether to log an MLflow artifact.
+                Default True.
+                If True, log runner.model as an MLflow artifact
+                for the current run.
+            interval (int): Logging interval (every k iterations).
+            ignore_last (bool): Ignore the log of last iterations in each epoch
+                if less than `interval`.
+            reset_flag (bool): Whether to clear the output buffer after logging
+            by_epoch (bool): Whether EpochBasedRunner is used.
+
+        .. _MLflow:
+            https://www.mlflow.org/docs/latest/index.html
+        """
+        super(MlflowLoggerHook, self).__init__(interval, ignore_last,
+                                               reset_flag, by_epoch)
+        self.import_mlflow()
+        self.exp_name = exp_name
+        self.tags = tags
+        self.log_model = log_model
+
+    def import_mlflow(self):
+        try:
+            import mlflow
+            import mlflow.pytorch as mlflow_pytorch
+        except ImportError:
+            raise ImportError(
+                'Please run "pip install mlflow" to install mlflow')
+        self.mlflow = mlflow
+        self.mlflow_pytorch = mlflow_pytorch
+
+    @master_only
+    def before_run(self, runner):
+        super(MlflowLoggerHook, self).before_run(runner)
+        if self.exp_name is not None:
+            self.mlflow.set_experiment(self.exp_name)
+        if self.tags is not None:
+            self.mlflow.set_tags(self.tags)
+
+    @master_only
+    def log(self, runner):
+        tags = self.get_loggable_tags(runner)
+        if tags:
+            self.mlflow.log_metrics(tags, step=self.get_iter(runner))
+
+    @master_only
+    def after_run(self, runner):
+        if self.log_model:
+            self.mlflow_pytorch.log_model(runner.model, 'models')
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/neptune.py b/annotator/uniformer/mmcv/runner/hooks/logger/neptune.py
new file mode 100644
index 0000000000000000000000000000000000000000..7a38772b0c93a8608f32c6357b8616e77c139dc9
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/neptune.py
@@ -0,0 +1,82 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from ...dist_utils import master_only
+from ..hook import HOOKS
+from .base import LoggerHook
+
+
+@HOOKS.register_module()
+class NeptuneLoggerHook(LoggerHook):
+    """Class to log metrics to NeptuneAI.
+
+    It requires `neptune-client` to be installed.
+
+    Args:
+        init_kwargs (dict): a dict contains the initialization keys as below:
+            - project (str): Name of a project in a form of
+                namespace/project_name. If None, the value of
+                NEPTUNE_PROJECT environment variable will be taken.
+            - api_token (str): User’s API token.
+                If None, the value of NEPTUNE_API_TOKEN environment
+                variable will be taken. Note: It is strongly recommended
+                to use NEPTUNE_API_TOKEN environment variable rather than
+                placing your API token in plain text in your source code.
+            - name (str, optional, default is 'Untitled'): Editable name of
+                the run. Name is displayed in the run's Details and in
+                Runs table as a column.
+            Check https://docs.neptune.ai/api-reference/neptune#init for
+                more init arguments.
+        interval (int): Logging interval (every k iterations).
+        ignore_last (bool): Ignore the log of last iterations in each epoch
+            if less than `interval`.
+        reset_flag (bool): Whether to clear the output buffer after logging
+        by_epoch (bool): Whether EpochBasedRunner is used.
+
+    .. _NeptuneAI:
+        https://docs.neptune.ai/you-should-know/logging-metadata
+    """
+
+    def __init__(self,
+                 init_kwargs=None,
+                 interval=10,
+                 ignore_last=True,
+                 reset_flag=True,
+                 with_step=True,
+                 by_epoch=True):
+
+        super(NeptuneLoggerHook, self).__init__(interval, ignore_last,
+                                                reset_flag, by_epoch)
+        self.import_neptune()
+        self.init_kwargs = init_kwargs
+        self.with_step = with_step
+
+    def import_neptune(self):
+        try:
+            import neptune.new as neptune
+        except ImportError:
+            raise ImportError(
+                'Please run "pip install neptune-client" to install neptune')
+        self.neptune = neptune
+        self.run = None
+
+    @master_only
+    def before_run(self, runner):
+        if self.init_kwargs:
+            self.run = self.neptune.init(**self.init_kwargs)
+        else:
+            self.run = self.neptune.init()
+
+    @master_only
+    def log(self, runner):
+        tags = self.get_loggable_tags(runner)
+        if tags:
+            for tag_name, tag_value in tags.items():
+                if self.with_step:
+                    self.run[tag_name].log(
+                        tag_value, step=self.get_iter(runner))
+                else:
+                    tags['global_step'] = self.get_iter(runner)
+                    self.run[tag_name].log(tags)
+
+    @master_only
+    def after_run(self, runner):
+        self.run.stop()
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/pavi.py b/annotator/uniformer/mmcv/runner/hooks/logger/pavi.py
new file mode 100644
index 0000000000000000000000000000000000000000..1dcf146d8163aff1363e9764999b0a74d674a595
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/pavi.py
@@ -0,0 +1,117 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import json
+import os
+import os.path as osp
+
+import torch
+import yaml
+
+import annotator.uniformer.mmcv as mmcv
+from ....parallel.utils import is_module_wrapper
+from ...dist_utils import master_only
+from ..hook import HOOKS
+from .base import LoggerHook
+
+
+@HOOKS.register_module()
+class PaviLoggerHook(LoggerHook):
+
+    def __init__(self,
+                 init_kwargs=None,
+                 add_graph=False,
+                 add_last_ckpt=False,
+                 interval=10,
+                 ignore_last=True,
+                 reset_flag=False,
+                 by_epoch=True,
+                 img_key='img_info'):
+        super(PaviLoggerHook, self).__init__(interval, ignore_last, reset_flag,
+                                             by_epoch)
+        self.init_kwargs = init_kwargs
+        self.add_graph = add_graph
+        self.add_last_ckpt = add_last_ckpt
+        self.img_key = img_key
+
+    @master_only
+    def before_run(self, runner):
+        super(PaviLoggerHook, self).before_run(runner)
+        try:
+            from pavi import SummaryWriter
+        except ImportError:
+            raise ImportError('Please run "pip install pavi" to install pavi.')
+
+        self.run_name = runner.work_dir.split('/')[-1]
+
+        if not self.init_kwargs:
+            self.init_kwargs = dict()
+        self.init_kwargs['name'] = self.run_name
+        self.init_kwargs['model'] = runner._model_name
+        if runner.meta is not None:
+            if 'config_dict' in runner.meta:
+                config_dict = runner.meta['config_dict']
+                assert isinstance(
+                    config_dict,
+                    dict), ('meta["config_dict"] has to be of a dict, '
+                            f'but got {type(config_dict)}')
+            elif 'config_file' in runner.meta:
+                config_file = runner.meta['config_file']
+                config_dict = dict(mmcv.Config.fromfile(config_file))
+            else:
+                config_dict = None
+            if config_dict is not None:
+                # 'max_.*iter' is parsed in pavi sdk as the maximum iterations
+                #  to properly set up the progress bar.
+                config_dict = config_dict.copy()
+                config_dict.setdefault('max_iter', runner.max_iters)
+                # non-serializable values are first converted in
+                # mmcv.dump to json
+                config_dict = json.loads(
+                    mmcv.dump(config_dict, file_format='json'))
+                session_text = yaml.dump(config_dict)
+                self.init_kwargs['session_text'] = session_text
+        self.writer = SummaryWriter(**self.init_kwargs)
+
+    def get_step(self, runner):
+        """Get the total training step/epoch."""
+        if self.get_mode(runner) == 'val' and self.by_epoch:
+            return self.get_epoch(runner)
+        else:
+            return self.get_iter(runner)
+
+    @master_only
+    def log(self, runner):
+        tags = self.get_loggable_tags(runner, add_mode=False)
+        if tags:
+            self.writer.add_scalars(
+                self.get_mode(runner), tags, self.get_step(runner))
+
+    @master_only
+    def after_run(self, runner):
+        if self.add_last_ckpt:
+            ckpt_path = osp.join(runner.work_dir, 'latest.pth')
+            if osp.islink(ckpt_path):
+                ckpt_path = osp.join(runner.work_dir, os.readlink(ckpt_path))
+
+            if osp.isfile(ckpt_path):
+                # runner.epoch += 1 has been done before `after_run`.
+                iteration = runner.epoch if self.by_epoch else runner.iter
+                return self.writer.add_snapshot_file(
+                    tag=self.run_name,
+                    snapshot_file_path=ckpt_path,
+                    iteration=iteration)
+
+        # flush the buffer and send a task ending signal to Pavi
+        self.writer.close()
+
+    @master_only
+    def before_epoch(self, runner):
+        if runner.epoch == 0 and self.add_graph:
+            if is_module_wrapper(runner.model):
+                _model = runner.model.module
+            else:
+                _model = runner.model
+            device = next(_model.parameters()).device
+            data = next(iter(runner.data_loader))
+            image = data[self.img_key][0:1].to(device)
+            with torch.no_grad():
+                self.writer.add_graph(_model, image)
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/tensorboard.py b/annotator/uniformer/mmcv/runner/hooks/logger/tensorboard.py
new file mode 100644
index 0000000000000000000000000000000000000000..4dd5011dc08def6c09eef86d3ce5b124c9fc5372
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/tensorboard.py
@@ -0,0 +1,57 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import os.path as osp
+
+from annotator.uniformer.mmcv.utils import TORCH_VERSION, digit_version
+from ...dist_utils import master_only
+from ..hook import HOOKS
+from .base import LoggerHook
+
+
+@HOOKS.register_module()
+class TensorboardLoggerHook(LoggerHook):
+
+    def __init__(self,
+                 log_dir=None,
+                 interval=10,
+                 ignore_last=True,
+                 reset_flag=False,
+                 by_epoch=True):
+        super(TensorboardLoggerHook, self).__init__(interval, ignore_last,
+                                                    reset_flag, by_epoch)
+        self.log_dir = log_dir
+
+    @master_only
+    def before_run(self, runner):
+        super(TensorboardLoggerHook, self).before_run(runner)
+        if (TORCH_VERSION == 'parrots'
+                or digit_version(TORCH_VERSION) < digit_version('1.1')):
+            try:
+                from tensorboardX import SummaryWriter
+            except ImportError:
+                raise ImportError('Please install tensorboardX to use '
+                                  'TensorboardLoggerHook.')
+        else:
+            try:
+                from torch.utils.tensorboard import SummaryWriter
+            except ImportError:
+                raise ImportError(
+                    'Please run "pip install future tensorboard" to install '
+                    'the dependencies to use torch.utils.tensorboard '
+                    '(applicable to PyTorch 1.1 or higher)')
+
+        if self.log_dir is None:
+            self.log_dir = osp.join(runner.work_dir, 'tf_logs')
+        self.writer = SummaryWriter(self.log_dir)
+
+    @master_only
+    def log(self, runner):
+        tags = self.get_loggable_tags(runner, allow_text=True)
+        for tag, val in tags.items():
+            if isinstance(val, str):
+                self.writer.add_text(tag, val, self.get_iter(runner))
+            else:
+                self.writer.add_scalar(tag, val, self.get_iter(runner))
+
+    @master_only
+    def after_run(self, runner):
+        self.writer.close()
diff --git a/annotator/uniformer/mmcv/runner/hooks/logger/text.py b/annotator/uniformer/mmcv/runner/hooks/logger/text.py
new file mode 100644
index 0000000000000000000000000000000000000000..87b1a3eca9595a130121526f8b4c29915387ab35
--- /dev/null
+++ b/annotator/uniformer/mmcv/runner/hooks/logger/text.py
@@ -0,0 +1,256 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import datetime
+import os
+import os.path as osp
+from collections import OrderedDict
+
+import torch
+import torch.distributed as dist
+
+import annotator.uniformer.mmcv as mmcv
+from annotator.uniformer.mmcv.fileio.file_client import FileClient
+from annotator.uniformer.mmcv.utils import is_tuple_of, scandir
+from ..hook import HOOKS
+from .base import LoggerHook
+
+
+@HOOKS.register_module()
+class TextLoggerHook(LoggerHook):
+    """Logger hook in text.
+
+    In this logger hook, the information will be printed on terminal and
+    saved in json file.
+
+    Args:
+        by_epoch (bool, optional): Whether EpochBasedRunner is used.
+            Default: True.
+        interval (int, optional): Logging interval (every k iterations).
+            Default: 10.
+        ignore_last (bool, optional): Ignore the log of last iterations in each
+            epoch if less than :attr:`interval`. Default: True.
+        reset_flag (bool, optional): Whether to clear the output buffer after
+            logging. Default: False.
+        interval_exp_name (int, optional): Logging interval for experiment
+            name. This feature is to help users conveniently get the experiment
+            information from screen or log file. Default: 1000.
+        out_dir (str, optional): Logs are saved in ``runner.work_dir`` default.
+            If ``out_dir`` is specified, logs will be copied to a new directory
+            which is the concatenation of ``out_dir`` and the last level
+            directory of ``runner.work_dir``. Default: None.
+            `New in version 1.3.16.`
+        out_suffix (str or tuple[str], optional): Those filenames ending with
+            ``out_suffix`` will be copied to ``out_dir``.
+            Default: ('.log.json', '.log', '.py').
+            `New in version 1.3.16.`
+        keep_local (bool, optional): Whether to keep local log when
+            :attr:`out_dir` is specified. If False, the local log will be
+            removed. Default: True.
+            `New in version 1.3.16.`
+        file_client_args (dict, optional): Arguments to instantiate a
+            FileClient. See :class:`mmcv.fileio.FileClient` for details.
+            Default: None.
+            `New in version 1.3.16.`
+    """
+
+    def __init__(self,
+                 by_epoch=True,
+                 interval=10,
+                 ignore_last=True,
+                 reset_flag=False,
+                 interval_exp_name=1000,
+                 out_dir=None,
+                 out_suffix=('.log.json', '.log', '.py'),
+                 keep_local=True,
+                 file_client_args=None):
+        super(TextLoggerHook, self).__init__(interval, ignore_last, reset_flag,
+                                             by_epoch)
+        self.by_epoch = by_epoch
+        self.time_sec_tot = 0
+        self.interval_exp_name = interval_exp_name
+
+        if out_dir is None and file_client_args is not None:
+            raise ValueError(
+                'file_client_args should be "None" when `out_dir` is not'
+                'specified.')
+        self.out_dir = out_dir
+
+        if not (out_dir is None or isinstance(out_dir, str)
+                or is_tuple_of(out_dir, str)):
+            raise TypeError('out_dir should be  "None" or string or tuple of '
+                            'string, but got {out_dir}')
+        self.out_suffix = out_suffix
+
+        self.keep_local = keep_local
+        self.file_client_args = file_client_args
+        if self.out_dir is not None:
+            self.file_client = FileClient.infer_client(file_client_args,
+                                                       self.out_dir)
+
+    def before_run(self, runner):
+        super(TextLoggerHook, self).before_run(runner)
+
+        if self.out_dir is not None:
+            self.file_client = FileClient.infer_client(self.file_client_args,
+                                                       self.out_dir)
+            # The final `self.out_dir` is the concatenation of `self.out_dir`
+            # and the last level directory of `runner.work_dir`
+            basename = osp.basename(runner.work_dir.rstrip(osp.sep))
+            self.out_dir = self.file_client.join_path(self.out_dir, basename)
+            runner.logger.info(
+                (f'Text logs will be saved to {self.out_dir} by '
+                 f'{self.file_client.name} after the training process.'))
+
+        self.start_iter = runner.iter
+        self.json_log_path = osp.join(runner.work_dir,
+                                      f'{runner.timestamp}.log.json')
+        if runner.meta is not None:
+            self._dump_log(runner.meta, runner)
+
+    def _get_max_memory(self, runner):
+        device = getattr(runner.model, 'output_device', None)
+        mem = torch.cuda.max_memory_allocated(device=device)
+        mem_mb = torch.tensor([mem / (1024 * 1024)],
+                              dtype=torch.int,
+                              device=device)
+        if runner.world_size > 1:
+            dist.reduce(mem_mb, 0, op=dist.ReduceOp.MAX)
+        return mem_mb.item()
+
+    def _log_info(self, log_dict, runner):
+        # print exp name for users to distinguish experiments
+        # at every ``interval_exp_name`` iterations and the end of each epoch
+        if runner.meta is not None and 'exp_name' in runner.meta:
+            if (self.every_n_iters(runner, self.interval_exp_name)) or (
+                    self.by_epoch and self.end_of_epoch(runner)):
+                exp_info = f'Exp name: {runner.meta["exp_name"]}'
+                runner.logger.info(exp_info)
+
+        if log_dict['mode'] == 'train':
+            if isinstance(log_dict['lr'], dict):
+                lr_str = []
+                for k, val in log_dict['lr'].items():
+                    lr_str.append(f'lr_{k}: {val:.3e}')
+                lr_str = ' '.join(lr_str)
+            else:
+                lr_str = f'lr: {log_dict["lr"]:.3e}'
+
+            # by epoch: Epoch [4][100/1000]
+            # by iter:  Iter [100/100000]
+            if self.by_epoch:
+                log_str = f'Epoch [{log_dict["epoch"]}]' \
+                          f'[{log_dict["iter"]}/{len(runner.data_loader)}]\t'
+            else:
+                log_str = f'Iter [{log_dict["iter"]}/{runner.max_iters}]\t'
+            log_str += f'{lr_str}, '
+
+            if 'time' in log_dict.keys():
+                self.time_sec_tot += (log_dict['time'] * self.interval)
+                time_sec_avg = self.time_sec_tot / (
+                    runner.iter - self.start_iter + 1)
+                eta_sec = time_