Upload 8 files

models/models/__init__.py

@@ -0,0 +1 @@
+# init

models/models/common.py

@@ -0,0 +1,2019 @@
+import math
+from copy import copy
+from pathlib import Path
+
+import numpy as np
+import pandas as pd
+import requests
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torchvision.ops import DeformConv2d
+from PIL import Image
+from torch.cuda import amp
+
+from utils.datasets import letterbox
+from utils.general import non_max_suppression, make_divisible, scale_coords, increment_path, xyxy2xywh
+from utils.plots import color_list, plot_one_box
+from utils.torch_utils import time_synchronized
+
+
+##### basic ####
+
+def autopad(k, p=None):  # kernel, padding
+    # Pad to 'same'
+    if p is None:
+        p = k // 2 if isinstance(k, int) else [x // 2 for x in k]  # auto-pad
+    return p
+
+
+class MP(nn.Module):
+    def __init__(self, k=2):
+        super(MP, self).__init__()
+        self.m = nn.MaxPool2d(kernel_size=k, stride=k)
+
+    def forward(self, x):
+        return self.m(x)
+
+
+class SP(nn.Module):
+    def __init__(self, k=3, s=1):
+        super(SP, self).__init__()
+        self.m = nn.MaxPool2d(kernel_size=k, stride=s, padding=k // 2)
+
+    def forward(self, x):
+        return self.m(x)
+    
+    
+class ReOrg(nn.Module):
+    def __init__(self):
+        super(ReOrg, self).__init__()
+
+    def forward(self, x):  # x(b,c,w,h) -> y(b,4c,w/2,h/2)
+        return torch.cat([x[..., ::2, ::2], x[..., 1::2, ::2], x[..., ::2, 1::2], x[..., 1::2, 1::2]], 1)
+
+
+class Concat(nn.Module):
+    def __init__(self, dimension=1):
+        super(Concat, self).__init__()
+        self.d = dimension
+
+    def forward(self, x):
+        return torch.cat(x, self.d)
+
+
+class Chuncat(nn.Module):
+    def __init__(self, dimension=1):
+        super(Chuncat, self).__init__()
+        self.d = dimension
+
+    def forward(self, x):
+        x1 = []
+        x2 = []
+        for xi in x:
+            xi1, xi2 = xi.chunk(2, self.d)
+            x1.append(xi1)
+            x2.append(xi2)
+        return torch.cat(x1+x2, self.d)
+
+
+class Shortcut(nn.Module):
+    def __init__(self, dimension=0):
+        super(Shortcut, self).__init__()
+        self.d = dimension
+
+    def forward(self, x):
+        return x[0]+x[1]
+
+
+class Foldcut(nn.Module):
+    def __init__(self, dimension=0):
+        super(Foldcut, self).__init__()
+        self.d = dimension
+
+    def forward(self, x):
+        x1, x2 = x.chunk(2, self.d)
+        return x1+x2
+
+
+class Conv(nn.Module):
+    # Standard convolution
+    def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True):  # ch_in, ch_out, kernel, stride, padding, groups
+        super(Conv, self).__init__()
+        self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=False)
+        self.bn = nn.BatchNorm2d(c2)
+        self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())
+
+    def forward(self, x):
+        return self.act(self.bn(self.conv(x)))
+
+    def fuseforward(self, x):
+        return self.act(self.conv(x))
+    
+
+class RobustConv(nn.Module):
+    # Robust convolution (use high kernel size 7-11 for: downsampling and other layers). Train for 300 - 450 epochs.
+    def __init__(self, c1, c2, k=7, s=1, p=None, g=1, act=True, layer_scale_init_value=1e-6):  # ch_in, ch_out, kernel, stride, padding, groups
+        super(RobustConv, self).__init__()
+        self.conv_dw = Conv(c1, c1, k=k, s=s, p=p, g=c1, act=act)
+        self.conv1x1 = nn.Conv2d(c1, c2, 1, 1, 0, groups=1, bias=True)
+        self.gamma = nn.Parameter(layer_scale_init_value * torch.ones(c2)) if layer_scale_init_value > 0 else None
+
+    def forward(self, x):
+        x = x.to(memory_format=torch.channels_last)
+        x = self.conv1x1(self.conv_dw(x))
+        if self.gamma is not None:
+            x = x.mul(self.gamma.reshape(1, -1, 1, 1)) 
+        return x
+
+
+class RobustConv2(nn.Module):
+    # Robust convolution 2 (use [32, 5, 2] or [32, 7, 4] or [32, 11, 8] for one of the paths in CSP).
+    def __init__(self, c1, c2, k=7, s=4, p=None, g=1, act=True, layer_scale_init_value=1e-6):  # ch_in, ch_out, kernel, stride, padding, groups
+        super(RobustConv2, self).__init__()
+        self.conv_strided = Conv(c1, c1, k=k, s=s, p=p, g=c1, act=act)
+        self.conv_deconv = nn.ConvTranspose2d(in_channels=c1, out_channels=c2, kernel_size=s, stride=s, 
+                                              padding=0, bias=True, dilation=1, groups=1
+        )
+        self.gamma = nn.Parameter(layer_scale_init_value * torch.ones(c2)) if layer_scale_init_value > 0 else None
+
+    def forward(self, x):
+        x = self.conv_deconv(self.conv_strided(x))
+        if self.gamma is not None:
+            x = x.mul(self.gamma.reshape(1, -1, 1, 1)) 
+        return x
+    
+
+def DWConv(c1, c2, k=1, s=1, act=True):
+    # Depthwise convolution
+    return Conv(c1, c2, k, s, g=math.gcd(c1, c2), act=act)
+
+
+class GhostConv(nn.Module):
+    # Ghost Convolution https://github.com/huawei-noah/ghostnet
+    def __init__(self, c1, c2, k=1, s=1, g=1, act=True):  # ch_in, ch_out, kernel, stride, groups
+        super(GhostConv, self).__init__()
+        c_ = c2 // 2  # hidden channels
+        self.cv1 = Conv(c1, c_, k, s, None, g, act)
+        self.cv2 = Conv(c_, c_, 5, 1, None, c_, act)
+
+    def forward(self, x):
+        y = self.cv1(x)
+        return torch.cat([y, self.cv2(y)], 1)
+
+
+class Stem(nn.Module):
+    # Stem
+    def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True):  # ch_in, ch_out, kernel, stride, padding, groups
+        super(Stem, self).__init__()
+        c_ = int(c2/2)  # hidden channels
+        self.cv1 = Conv(c1, c_, 3, 2)
+        self.cv2 = Conv(c_, c_, 1, 1)
+        self.cv3 = Conv(c_, c_, 3, 2)
+        self.pool = torch.nn.MaxPool2d(2, stride=2)
+        self.cv4 = Conv(2 * c_, c2, 1, 1)
+
+    def forward(self, x):
+        x = self.cv1(x)
+        return self.cv4(torch.cat((self.cv3(self.cv2(x)), self.pool(x)), dim=1))
+
+
+class DownC(nn.Module):
+    # Spatial pyramid pooling layer used in YOLOv3-SPP
+    def __init__(self, c1, c2, n=1, k=2):
+        super(DownC, self).__init__()
+        c_ = int(c1)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_, c2//2, 3, k)
+        self.cv3 = Conv(c1, c2//2, 1, 1)
+        self.mp = nn.MaxPool2d(kernel_size=k, stride=k)
+
+    def forward(self, x):
+        return torch.cat((self.cv2(self.cv1(x)), self.cv3(self.mp(x))), dim=1)
+
+
+class SPP(nn.Module):
+    # Spatial pyramid pooling layer used in YOLOv3-SPP
+    def __init__(self, c1, c2, k=(5, 9, 13)):
+        super(SPP, self).__init__()
+        c_ = c1 // 2  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_ * (len(k) + 1), c2, 1, 1)
+        self.m = nn.ModuleList([nn.MaxPool2d(kernel_size=x, stride=1, padding=x // 2) for x in k])
+
+    def forward(self, x):
+        x = self.cv1(x)
+        return self.cv2(torch.cat([x] + [m(x) for m in self.m], 1))
+    
+
+class Bottleneck(nn.Module):
+    # Darknet bottleneck
+    def __init__(self, c1, c2, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, shortcut, groups, expansion
+        super(Bottleneck, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_, c2, 3, 1, g=g)
+        self.add = shortcut and c1 == c2
+
+    def forward(self, x):
+        return x + self.cv2(self.cv1(x)) if self.add else self.cv2(self.cv1(x))
+
+
+class Res(nn.Module):
+    # ResNet bottleneck
+    def __init__(self, c1, c2, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, shortcut, groups, expansion
+        super(Res, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_, c_, 3, 1, g=g)
+        self.cv3 = Conv(c_, c2, 1, 1)
+        self.add = shortcut and c1 == c2
+
+    def forward(self, x):
+        return x + self.cv3(self.cv2(self.cv1(x))) if self.add else self.cv3(self.cv2(self.cv1(x)))
+
+
+class ResX(Res):
+    # ResNet bottleneck
+    def __init__(self, c1, c2, shortcut=True, g=32, e=0.5):  # ch_in, ch_out, shortcut, groups, expansion
+        super().__init__(c1, c2, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+
+
+class Ghost(nn.Module):
+    # Ghost Bottleneck https://github.com/huawei-noah/ghostnet
+    def __init__(self, c1, c2, k=3, s=1):  # ch_in, ch_out, kernel, stride
+        super(Ghost, self).__init__()
+        c_ = c2 // 2
+        self.conv = nn.Sequential(GhostConv(c1, c_, 1, 1),  # pw
+                                  DWConv(c_, c_, k, s, act=False) if s == 2 else nn.Identity(),  # dw
+                                  GhostConv(c_, c2, 1, 1, act=False))  # pw-linear
+        self.shortcut = nn.Sequential(DWConv(c1, c1, k, s, act=False),
+                                      Conv(c1, c2, 1, 1, act=False)) if s == 2 else nn.Identity()
+
+    def forward(self, x):
+        return self.conv(x) + self.shortcut(x)
+
+##### end of basic #####
+
+
+##### cspnet #####
+
+class SPPCSPC(nn.Module):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5, k=(5, 9, 13)):
+        super(SPPCSPC, self).__init__()
+        c_ = int(2 * c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c1, c_, 1, 1)
+        self.cv3 = Conv(c_, c_, 3, 1)
+        self.cv4 = Conv(c_, c_, 1, 1)
+        self.m = nn.ModuleList([nn.MaxPool2d(kernel_size=x, stride=1, padding=x // 2) for x in k])
+        self.cv5 = Conv(4 * c_, c_, 1, 1)
+        self.cv6 = Conv(c_, c_, 3, 1)
+        self.cv7 = Conv(2 * c_, c2, 1, 1)
+
+    def forward(self, x):
+        x1 = self.cv4(self.cv3(self.cv1(x)))
+        y1 = self.cv6(self.cv5(torch.cat([x1] + [m(x1) for m in self.m], 1)))
+        y2 = self.cv2(x)
+        return self.cv7(torch.cat((y1, y2), dim=1))
+
+class GhostSPPCSPC(SPPCSPC):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5, k=(5, 9, 13)):
+        super().__init__(c1, c2, n, shortcut, g, e, k)
+        c_ = int(2 * c2 * e)  # hidden channels
+        self.cv1 = GhostConv(c1, c_, 1, 1)
+        self.cv2 = GhostConv(c1, c_, 1, 1)
+        self.cv3 = GhostConv(c_, c_, 3, 1)
+        self.cv4 = GhostConv(c_, c_, 1, 1)
+        self.cv5 = GhostConv(4 * c_, c_, 1, 1)
+        self.cv6 = GhostConv(c_, c_, 3, 1)
+        self.cv7 = GhostConv(2 * c_, c2, 1, 1)
+
+
+class GhostStem(Stem):
+    # Stem
+    def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True):  # ch_in, ch_out, kernel, stride, padding, groups
+        super().__init__(c1, c2, k, s, p, g, act)
+        c_ = int(c2/2)  # hidden channels
+        self.cv1 = GhostConv(c1, c_, 3, 2)
+        self.cv2 = GhostConv(c_, c_, 1, 1)
+        self.cv3 = GhostConv(c_, c_, 3, 2)
+        self.cv4 = GhostConv(2 * c_, c2, 1, 1)
+        
+
+class BottleneckCSPA(nn.Module):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(BottleneckCSPA, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c1, c_, 1, 1)
+        self.cv3 = Conv(2 * c_, c2, 1, 1)
+        self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        y1 = self.m(self.cv1(x))
+        y2 = self.cv2(x)
+        return self.cv3(torch.cat((y1, y2), dim=1))
+
+
+class BottleneckCSPB(nn.Module):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(BottleneckCSPB, self).__init__()
+        c_ = int(c2)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_, c_, 1, 1)
+        self.cv3 = Conv(2 * c_, c2, 1, 1)
+        self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        x1 = self.cv1(x)
+        y1 = self.m(x1)
+        y2 = self.cv2(x1)
+        return self.cv3(torch.cat((y1, y2), dim=1))
+
+
+class BottleneckCSPC(nn.Module):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(BottleneckCSPC, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c1, c_, 1, 1)
+        self.cv3 = Conv(c_, c_, 1, 1)
+        self.cv4 = Conv(2 * c_, c2, 1, 1)
+        self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        y1 = self.cv3(self.m(self.cv1(x)))
+        y2 = self.cv2(x)
+        return self.cv4(torch.cat((y1, y2), dim=1))
+
+
+class ResCSPA(BottleneckCSPA):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[Res(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class ResCSPB(BottleneckCSPB):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2)  # hidden channels
+        self.m = nn.Sequential(*[Res(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class ResCSPC(BottleneckCSPC):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[Res(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class ResXCSPA(ResCSPA):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=32, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[Res(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+
+class ResXCSPB(ResCSPB):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=32, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2)  # hidden channels
+        self.m = nn.Sequential(*[Res(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+
+class ResXCSPC(ResCSPC):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=32, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[Res(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+
+class GhostCSPA(BottleneckCSPA):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[Ghost(c_, c_) for _ in range(n)])
+
+
+class GhostCSPB(BottleneckCSPB):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2)  # hidden channels
+        self.m = nn.Sequential(*[Ghost(c_, c_) for _ in range(n)])
+
+
+class GhostCSPC(BottleneckCSPC):
+    # CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[Ghost(c_, c_) for _ in range(n)])
+
+##### end of cspnet #####
+
+
+##### yolor #####
+
+class ImplicitA(nn.Module):
+    def __init__(self, channel, mean=0., std=.02):
+        super(ImplicitA, self).__init__()
+        self.channel = channel
+        self.mean = mean
+        self.std = std
+        self.implicit = nn.Parameter(torch.zeros(1, channel, 1, 1))
+        nn.init.normal_(self.implicit, mean=self.mean, std=self.std)
+
+    def forward(self, x):
+        return self.implicit + x
+    
+
+class ImplicitM(nn.Module):
+    def __init__(self, channel, mean=1., std=.02):
+        super(ImplicitM, self).__init__()
+        self.channel = channel
+        self.mean = mean
+        self.std = std
+        self.implicit = nn.Parameter(torch.ones(1, channel, 1, 1))
+        nn.init.normal_(self.implicit, mean=self.mean, std=self.std)
+
+    def forward(self, x):
+        return self.implicit * x
+    
+##### end of yolor #####
+
+
+##### repvgg #####
+
+class RepConv(nn.Module):
+    # Represented convolution
+    # https://arxiv.org/abs/2101.03697
+
+    def __init__(self, c1, c2, k=3, s=1, p=None, g=1, act=True, deploy=False):
+        super(RepConv, self).__init__()
+
+        self.deploy = deploy
+        self.groups = g
+        self.in_channels = c1
+        self.out_channels = c2
+
+        assert k == 3
+        assert autopad(k, p) == 1
+
+        padding_11 = autopad(k, p) - k // 2
+
+        self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())
+
+        if deploy:
+            self.rbr_reparam = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=True)
+
+        else:
+            self.rbr_identity = (nn.BatchNorm2d(num_features=c1) if c2 == c1 and s == 1 else None)
+
+            self.rbr_dense = nn.Sequential(
+                nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=False),
+                nn.BatchNorm2d(num_features=c2),
+            )
+
+            self.rbr_1x1 = nn.Sequential(
+                nn.Conv2d( c1, c2, 1, s, padding_11, groups=g, bias=False),
+                nn.BatchNorm2d(num_features=c2),
+            )
+
+    def forward(self, inputs):
+        if hasattr(self, "rbr_reparam"):
+            return self.act(self.rbr_reparam(inputs))
+
+        if self.rbr_identity is None:
+            id_out = 0
+        else:
+            id_out = self.rbr_identity(inputs)
+
+        return self.act(self.rbr_dense(inputs) + self.rbr_1x1(inputs) + id_out)
+    
+    def get_equivalent_kernel_bias(self):
+        kernel3x3, bias3x3 = self._fuse_bn_tensor(self.rbr_dense)
+        kernel1x1, bias1x1 = self._fuse_bn_tensor(self.rbr_1x1)
+        kernelid, biasid = self._fuse_bn_tensor(self.rbr_identity)
+        return (
+            kernel3x3 + self._pad_1x1_to_3x3_tensor(kernel1x1) + kernelid,
+            bias3x3 + bias1x1 + biasid,
+        )
+
+    def _pad_1x1_to_3x3_tensor(self, kernel1x1):
+        if kernel1x1 is None:
+            return 0
+        else:
+            return nn.functional.pad(kernel1x1, [1, 1, 1, 1])
+
+    def _fuse_bn_tensor(self, branch):
+        if branch is None:
+            return 0, 0
+        if isinstance(branch, nn.Sequential):
+            kernel = branch[0].weight
+            running_mean = branch[1].running_mean
+            running_var = branch[1].running_var
+            gamma = branch[1].weight
+            beta = branch[1].bias
+            eps = branch[1].eps
+        else:
+            assert isinstance(branch, nn.BatchNorm2d)
+            if not hasattr(self, "id_tensor"):
+                input_dim = self.in_channels // self.groups
+                kernel_value = np.zeros(
+                    (self.in_channels, input_dim, 3, 3), dtype=np.float32
+                )
+                for i in range(self.in_channels):
+                    kernel_value[i, i % input_dim, 1, 1] = 1
+                self.id_tensor = torch.from_numpy(kernel_value).to(branch.weight.device)
+            kernel = self.id_tensor
+            running_mean = branch.running_mean
+            running_var = branch.running_var
+            gamma = branch.weight
+            beta = branch.bias
+            eps = branch.eps
+        std = (running_var + eps).sqrt()
+        t = (gamma / std).reshape(-1, 1, 1, 1)
+        return kernel * t, beta - running_mean * gamma / std
+
+    def repvgg_convert(self):
+        kernel, bias = self.get_equivalent_kernel_bias()
+        return (
+            kernel.detach().cpu().numpy(),
+            bias.detach().cpu().numpy(),
+        )
+
+    def fuse_conv_bn(self, conv, bn):
+
+        std = (bn.running_var + bn.eps).sqrt()
+        bias = bn.bias - bn.running_mean * bn.weight / std
+
+        t = (bn.weight / std).reshape(-1, 1, 1, 1)
+        weights = conv.weight * t
+
+        bn = nn.Identity()
+        conv = nn.Conv2d(in_channels = conv.in_channels,
+                              out_channels = conv.out_channels,
+                              kernel_size = conv.kernel_size,
+                              stride=conv.stride,
+                              padding = conv.padding,
+                              dilation = conv.dilation,
+                              groups = conv.groups,
+                              bias = True,
+                              padding_mode = conv.padding_mode)
+
+        conv.weight = torch.nn.Parameter(weights)
+        conv.bias = torch.nn.Parameter(bias)
+        return conv
+
+    def fuse_repvgg_block(self):    
+        if self.deploy:
+            return
+        print(f"RepConv.fuse_repvgg_block")
+                
+        self.rbr_dense = self.fuse_conv_bn(self.rbr_dense[0], self.rbr_dense[1])
+        
+        self.rbr_1x1 = self.fuse_conv_bn(self.rbr_1x1[0], self.rbr_1x1[1])
+        rbr_1x1_bias = self.rbr_1x1.bias
+        weight_1x1_expanded = torch.nn.functional.pad(self.rbr_1x1.weight, [1, 1, 1, 1])
+        
+        # Fuse self.rbr_identity
+        if (isinstance(self.rbr_identity, nn.BatchNorm2d) or isinstance(self.rbr_identity, nn.modules.batchnorm.SyncBatchNorm)):
+            # print(f"fuse: rbr_identity == BatchNorm2d or SyncBatchNorm")
+            identity_conv_1x1 = nn.Conv2d(
+                    in_channels=self.in_channels,
+                    out_channels=self.out_channels,
+                    kernel_size=1,
+                    stride=1,
+                    padding=0,
+                    groups=self.groups, 
+                    bias=False)
+            identity_conv_1x1.weight.data = identity_conv_1x1.weight.data.to(self.rbr_1x1.weight.data.device)
+            identity_conv_1x1.weight.data = identity_conv_1x1.weight.data.squeeze().squeeze()
+            # print(f" identity_conv_1x1.weight = {identity_conv_1x1.weight.shape}")
+            identity_conv_1x1.weight.data.fill_(0.0)
+            identity_conv_1x1.weight.data.fill_diagonal_(1.0)
+            identity_conv_1x1.weight.data = identity_conv_1x1.weight.data.unsqueeze(2).unsqueeze(3)
+            # print(f" identity_conv_1x1.weight = {identity_conv_1x1.weight.shape}")
+
+            identity_conv_1x1 = self.fuse_conv_bn(identity_conv_1x1, self.rbr_identity)
+            bias_identity_expanded = identity_conv_1x1.bias
+            weight_identity_expanded = torch.nn.functional.pad(identity_conv_1x1.weight, [1, 1, 1, 1])            
+        else:
+            # print(f"fuse: rbr_identity != BatchNorm2d, rbr_identity = {self.rbr_identity}")
+            bias_identity_expanded = torch.nn.Parameter( torch.zeros_like(rbr_1x1_bias) )
+            weight_identity_expanded = torch.nn.Parameter( torch.zeros_like(weight_1x1_expanded) )            
+        
+
+        #print(f"self.rbr_1x1.weight = {self.rbr_1x1.weight.shape}, ")
+        #print(f"weight_1x1_expanded = {weight_1x1_expanded.shape}, ")
+        #print(f"self.rbr_dense.weight = {self.rbr_dense.weight.shape}, ")
+
+        self.rbr_dense.weight = torch.nn.Parameter(self.rbr_dense.weight + weight_1x1_expanded + weight_identity_expanded)
+        self.rbr_dense.bias = torch.nn.Parameter(self.rbr_dense.bias + rbr_1x1_bias + bias_identity_expanded)
+                
+        self.rbr_reparam = self.rbr_dense
+        self.deploy = True
+
+        if self.rbr_identity is not None:
+            del self.rbr_identity
+            self.rbr_identity = None
+
+        if self.rbr_1x1 is not None:
+            del self.rbr_1x1
+            self.rbr_1x1 = None
+
+        if self.rbr_dense is not None:
+            del self.rbr_dense
+            self.rbr_dense = None
+
+
+class RepBottleneck(Bottleneck):
+    # Standard bottleneck
+    def __init__(self, c1, c2, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, shortcut, groups, expansion
+        super().__init__(c1, c2, shortcut=True, g=1, e=0.5)
+        c_ = int(c2 * e)  # hidden channels
+        self.cv2 = RepConv(c_, c2, 3, 1, g=g)
+
+
+class RepBottleneckCSPA(BottleneckCSPA):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[RepBottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+
+class RepBottleneckCSPB(BottleneckCSPB):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2)  # hidden channels
+        self.m = nn.Sequential(*[RepBottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+
+class RepBottleneckCSPC(BottleneckCSPC):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[RepBottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+
+class RepRes(Res):
+    # Standard bottleneck
+    def __init__(self, c1, c2, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, shortcut, groups, expansion
+        super().__init__(c1, c2, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.cv2 = RepConv(c_, c_, 3, 1, g=g)
+
+
+class RepResCSPA(ResCSPA):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[RepRes(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class RepResCSPB(ResCSPB):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2)  # hidden channels
+        self.m = nn.Sequential(*[RepRes(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class RepResCSPC(ResCSPC):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[RepRes(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class RepResX(ResX):
+    # Standard bottleneck
+    def __init__(self, c1, c2, shortcut=True, g=32, e=0.5):  # ch_in, ch_out, shortcut, groups, expansion
+        super().__init__(c1, c2, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.cv2 = RepConv(c_, c_, 3, 1, g=g)
+
+
+class RepResXCSPA(ResXCSPA):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=32, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[RepResX(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class RepResXCSPB(ResXCSPB):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=32, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2)  # hidden channels
+        self.m = nn.Sequential(*[RepResX(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+
+class RepResXCSPC(ResXCSPC):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=32, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super().__init__(c1, c2, n, shortcut, g, e)
+        c_ = int(c2 * e)  # hidden channels
+        self.m = nn.Sequential(*[RepResX(c_, c_, shortcut, g, e=0.5) for _ in range(n)])
+
+##### end of repvgg #####
+
+
+##### transformer #####
+
+class TransformerLayer(nn.Module):
+    # Transformer layer https://arxiv.org/abs/2010.11929 (LayerNorm layers removed for better performance)
+    def __init__(self, c, num_heads):
+        super().__init__()
+        self.q = nn.Linear(c, c, bias=False)
+        self.k = nn.Linear(c, c, bias=False)
+        self.v = nn.Linear(c, c, bias=False)
+        self.ma = nn.MultiheadAttention(embed_dim=c, num_heads=num_heads)
+        self.fc1 = nn.Linear(c, c, bias=False)
+        self.fc2 = nn.Linear(c, c, bias=False)
+
+    def forward(self, x):
+        x = self.ma(self.q(x), self.k(x), self.v(x))[0] + x
+        x = self.fc2(self.fc1(x)) + x
+        return x
+
+
+class TransformerBlock(nn.Module):
+    # Vision Transformer https://arxiv.org/abs/2010.11929
+    def __init__(self, c1, c2, num_heads, num_layers):
+        super().__init__()
+        self.conv = None
+        if c1 != c2:
+            self.conv = Conv(c1, c2)
+        self.linear = nn.Linear(c2, c2)  # learnable position embedding
+        self.tr = nn.Sequential(*[TransformerLayer(c2, num_heads) for _ in range(num_layers)])
+        self.c2 = c2
+
+    def forward(self, x):
+        if self.conv is not None:
+            x = self.conv(x)
+        b, _, w, h = x.shape
+        p = x.flatten(2)
+        p = p.unsqueeze(0)
+        p = p.transpose(0, 3)
+        p = p.squeeze(3)
+        e = self.linear(p)
+        x = p + e
+
+        x = self.tr(x)
+        x = x.unsqueeze(3)
+        x = x.transpose(0, 3)
+        x = x.reshape(b, self.c2, w, h)
+        return x
+
+##### end of transformer #####
+
+
+##### yolov5 #####
+
+class Focus(nn.Module):
+    # Focus wh information into c-space
+    def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True):  # ch_in, ch_out, kernel, stride, padding, groups
+        super(Focus, self).__init__()
+        self.conv = Conv(c1 * 4, c2, k, s, p, g, act)
+        # self.contract = Contract(gain=2)
+
+    def forward(self, x):  # x(b,c,w,h) -> y(b,4c,w/2,h/2)
+        return self.conv(torch.cat([x[..., ::2, ::2], x[..., 1::2, ::2], x[..., ::2, 1::2], x[..., 1::2, 1::2]], 1))
+        # return self.conv(self.contract(x))
+        
+
+class SPPF(nn.Module):
+    # Spatial Pyramid Pooling - Fast (SPPF) layer for YOLOv5 by Glenn Jocher
+    def __init__(self, c1, c2, k=5):  # equivalent to SPP(k=(5, 9, 13))
+        super().__init__()
+        c_ = c1 // 2  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_ * 4, c2, 1, 1)
+        self.m = nn.MaxPool2d(kernel_size=k, stride=1, padding=k // 2)
+
+    def forward(self, x):
+        x = self.cv1(x)
+        y1 = self.m(x)
+        y2 = self.m(y1)
+        return self.cv2(torch.cat([x, y1, y2, self.m(y2)], 1))
+    
+    
+class Contract(nn.Module):
+    # Contract width-height into channels, i.e. x(1,64,80,80) to x(1,256,40,40)
+    def __init__(self, gain=2):
+        super().__init__()
+        self.gain = gain
+
+    def forward(self, x):
+        N, C, H, W = x.size()  # assert (H / s == 0) and (W / s == 0), 'Indivisible gain'
+        s = self.gain
+        x = x.view(N, C, H // s, s, W // s, s)  # x(1,64,40,2,40,2)
+        x = x.permute(0, 3, 5, 1, 2, 4).contiguous()  # x(1,2,2,64,40,40)
+        return x.view(N, C * s * s, H // s, W // s)  # x(1,256,40,40)
+
+
+class Expand(nn.Module):
+    # Expand channels into width-height, i.e. x(1,64,80,80) to x(1,16,160,160)
+    def __init__(self, gain=2):
+        super().__init__()
+        self.gain = gain
+
+    def forward(self, x):
+        N, C, H, W = x.size()  # assert C / s ** 2 == 0, 'Indivisible gain'
+        s = self.gain
+        x = x.view(N, s, s, C // s ** 2, H, W)  # x(1,2,2,16,80,80)
+        x = x.permute(0, 3, 4, 1, 5, 2).contiguous()  # x(1,16,80,2,80,2)
+        return x.view(N, C // s ** 2, H * s, W * s)  # x(1,16,160,160)
+
+
+class NMS(nn.Module):
+    # Non-Maximum Suppression (NMS) module
+    conf = 0.25  # confidence threshold
+    iou = 0.45  # IoU threshold
+    classes = None  # (optional list) filter by class
+
+    def __init__(self):
+        super(NMS, self).__init__()
+
+    def forward(self, x):
+        return non_max_suppression(x[0], conf_thres=self.conf, iou_thres=self.iou, classes=self.classes)
+
+
+class autoShape(nn.Module):
+    # input-robust model wrapper for passing cv2/np/PIL/torch inputs. Includes preprocessing, inference and NMS
+    conf = 0.25  # NMS confidence threshold
+    iou = 0.45  # NMS IoU threshold
+    classes = None  # (optional list) filter by class
+
+    def __init__(self, model):
+        super(autoShape, self).__init__()
+        self.model = model.eval()
+
+    def autoshape(self):
+        print('autoShape already enabled, skipping... ')  # model already converted to model.autoshape()
+        return self
+
+    @torch.no_grad()
+    def forward(self, imgs, size=640, augment=False, profile=False):
+        # Inference from various sources. For height=640, width=1280, RGB images example inputs are:
+        #   filename:   imgs = 'data/samples/zidane.jpg'
+        #   URI:             = 'https://github.com/ultralytics/yolov5/releases/download/v1.0/zidane.jpg'
+        #   OpenCV:          = cv2.imread('image.jpg')[:,:,::-1]  # HWC BGR to RGB x(640,1280,3)
+        #   PIL:             = Image.open('image.jpg')  # HWC x(640,1280,3)
+        #   numpy:           = np.zeros((640,1280,3))  # HWC
+        #   torch:           = torch.zeros(16,3,320,640)  # BCHW (scaled to size=640, 0-1 values)
+        #   multiple:        = [Image.open('image1.jpg'), Image.open('image2.jpg'), ...]  # list of images
+
+        t = [time_synchronized()]
+        p = next(self.model.parameters())  # for device and type
+        if isinstance(imgs, torch.Tensor):  # torch
+            with amp.autocast(enabled=p.device.type != 'cpu'):
+                return self.model(imgs.to(p.device).type_as(p), augment, profile)  # inference
+
+        # Pre-process
+        n, imgs = (len(imgs), imgs) if isinstance(imgs, list) else (1, [imgs])  # number of images, list of images
+        shape0, shape1, files = [], [], []  # image and inference shapes, filenames
+        for i, im in enumerate(imgs):
+            f = f'image{i}'  # filename
+            if isinstance(im, str):  # filename or uri
+                im, f = np.asarray(Image.open(requests.get(im, stream=True).raw if im.startswith('http') else im)), im
+            elif isinstance(im, Image.Image):  # PIL Image
+                im, f = np.asarray(im), getattr(im, 'filename', f) or f
+            files.append(Path(f).with_suffix('.jpg').name)
+            if im.shape[0] < 5:  # image in CHW
+                im = im.transpose((1, 2, 0))  # reverse dataloader .transpose(2, 0, 1)
+            im = im[:, :, :3] if im.ndim == 3 else np.tile(im[:, :, None], 3)  # enforce 3ch input
+            s = im.shape[:2]  # HWC
+            shape0.append(s)  # image shape
+            g = (size / max(s))  # gain
+            shape1.append([y * g for y in s])
+            imgs[i] = im  # update
+        shape1 = [make_divisible(x, int(self.stride.max())) for x in np.stack(shape1, 0).max(0)]  # inference shape
+        x = [letterbox(im, new_shape=shape1, auto=False)[0] for im in imgs]  # pad
+        x = np.stack(x, 0) if n > 1 else x[0][None]  # stack
+        x = np.ascontiguousarray(x.transpose((0, 3, 1, 2)))  # BHWC to BCHW
+        x = torch.from_numpy(x).to(p.device).type_as(p) / 255.  # uint8 to fp16/32
+        t.append(time_synchronized())
+
+        with amp.autocast(enabled=p.device.type != 'cpu'):
+            # Inference
+            y = self.model(x, augment, profile)[0]  # forward
+            t.append(time_synchronized())
+
+            # Post-process
+            y = non_max_suppression(y, conf_thres=self.conf, iou_thres=self.iou, classes=self.classes)  # NMS
+            for i in range(n):
+                scale_coords(shape1, y[i][:, :4], shape0[i])
+
+            t.append(time_synchronized())
+            return Detections(imgs, y, files, t, self.names, x.shape)
+
+
+class Detections:
+    # detections class for YOLOv5 inference results
+    def __init__(self, imgs, pred, files, times=None, names=None, shape=None):
+        super(Detections, self).__init__()
+        d = pred[0].device  # device
+        gn = [torch.tensor([*[im.shape[i] for i in [1, 0, 1, 0]], 1., 1.], device=d) for im in imgs]  # normalizations
+        self.imgs = imgs  # list of images as numpy arrays
+        self.pred = pred  # list of tensors pred[0] = (xyxy, conf, cls)
+        self.names = names  # class names
+        self.files = files  # image filenames
+        self.xyxy = pred  # xyxy pixels
+        self.xywh = [xyxy2xywh(x) for x in pred]  # xywh pixels
+        self.xyxyn = [x / g for x, g in zip(self.xyxy, gn)]  # xyxy normalized
+        self.xywhn = [x / g for x, g in zip(self.xywh, gn)]  # xywh normalized
+        self.n = len(self.pred)  # number of images (batch size)
+        self.t = tuple((times[i + 1] - times[i]) * 1000 / self.n for i in range(3))  # timestamps (ms)
+        self.s = shape  # inference BCHW shape
+
+    def display(self, pprint=False, show=False, save=False, render=False, save_dir=''):
+        colors = color_list()
+        for i, (img, pred) in enumerate(zip(self.imgs, self.pred)):
+            str = f'image {i + 1}/{len(self.pred)}: {img.shape[0]}x{img.shape[1]} '
+            if pred is not None:
+                for c in pred[:, -1].unique():
+                    n = (pred[:, -1] == c).sum()  # detections per class
+                    str += f"{n} {self.names[int(c)]}{'s' * (n > 1)}, "  # add to string
+                if show or save or render:
+                    for *box, conf, cls in pred:  # xyxy, confidence, class
+                        label = f'{self.names[int(cls)]} {conf:.2f}'
+                        plot_one_box(box, img, label=label, color=colors[int(cls) % 10])
+            img = Image.fromarray(img.astype(np.uint8)) if isinstance(img, np.ndarray) else img  # from np
+            if pprint:
+                print(str.rstrip(', '))
+            if show:
+                img.show(self.files[i])  # show
+            if save:
+                f = self.files[i]
+                img.save(Path(save_dir) / f)  # save
+                print(f"{'Saved' * (i == 0)} {f}", end=',' if i < self.n - 1 else f' to {save_dir}\n')
+            if render:
+                self.imgs[i] = np.asarray(img)
+
+    def print(self):
+        self.display(pprint=True)  # print results
+        print(f'Speed: %.1fms pre-process, %.1fms inference, %.1fms NMS per image at shape {tuple(self.s)}' % self.t)
+
+    def show(self):
+        self.display(show=True)  # show results
+
+    def save(self, save_dir='runs/hub/exp'):
+        save_dir = increment_path(save_dir, exist_ok=save_dir != 'runs/hub/exp')  # increment save_dir
+        Path(save_dir).mkdir(parents=True, exist_ok=True)
+        self.display(save=True, save_dir=save_dir)  # save results
+
+    def render(self):
+        self.display(render=True)  # render results
+        return self.imgs
+
+    def pandas(self):
+        # return detections as pandas DataFrames, i.e. print(results.pandas().xyxy[0])
+        new = copy(self)  # return copy
+        ca = 'xmin', 'ymin', 'xmax', 'ymax', 'confidence', 'class', 'name'  # xyxy columns
+        cb = 'xcenter', 'ycenter', 'width', 'height', 'confidence', 'class', 'name'  # xywh columns
+        for k, c in zip(['xyxy', 'xyxyn', 'xywh', 'xywhn'], [ca, ca, cb, cb]):
+            a = [[x[:5] + [int(x[5]), self.names[int(x[5])]] for x in x.tolist()] for x in getattr(self, k)]  # update
+            setattr(new, k, [pd.DataFrame(x, columns=c) for x in a])
+        return new
+
+    def tolist(self):
+        # return a list of Detections objects, i.e. 'for result in results.tolist():'
+        x = [Detections([self.imgs[i]], [self.pred[i]], self.names, self.s) for i in range(self.n)]
+        for d in x:
+            for k in ['imgs', 'pred', 'xyxy', 'xyxyn', 'xywh', 'xywhn']:
+                setattr(d, k, getattr(d, k)[0])  # pop out of list
+        return x
+
+    def __len__(self):
+        return self.n
+
+
+class Classify(nn.Module):
+    # Classification head, i.e. x(b,c1,20,20) to x(b,c2)
+    def __init__(self, c1, c2, k=1, s=1, p=None, g=1):  # ch_in, ch_out, kernel, stride, padding, groups
+        super(Classify, self).__init__()
+        self.aap = nn.AdaptiveAvgPool2d(1)  # to x(b,c1,1,1)
+        self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g)  # to x(b,c2,1,1)
+        self.flat = nn.Flatten()
+
+    def forward(self, x):
+        z = torch.cat([self.aap(y) for y in (x if isinstance(x, list) else [x])], 1)  # cat if list
+        return self.flat(self.conv(z))  # flatten to x(b,c2)
+
+##### end of yolov5 ######
+
+
+##### orepa #####
+
+def transI_fusebn(kernel, bn):
+    gamma = bn.weight
+    std = (bn.running_var + bn.eps).sqrt()
+    return kernel * ((gamma / std).reshape(-1, 1, 1, 1)), bn.bias - bn.running_mean * gamma / std
+    
+    
+class ConvBN(nn.Module):
+    def __init__(self, in_channels, out_channels, kernel_size,
+                             stride=1, padding=0, dilation=1, groups=1, deploy=False, nonlinear=None):
+        super().__init__()
+        if nonlinear is None:
+            self.nonlinear = nn.Identity()
+        else:
+            self.nonlinear = nonlinear
+        if deploy:
+            self.conv = nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=kernel_size,
+                                      stride=stride, padding=padding, dilation=dilation, groups=groups, bias=True)
+        else:
+            self.conv = nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=kernel_size,
+                                            stride=stride, padding=padding, dilation=dilation, groups=groups, bias=False)
+            self.bn = nn.BatchNorm2d(num_features=out_channels)
+
+    def forward(self, x):
+        if hasattr(self, 'bn'):
+            return self.nonlinear(self.bn(self.conv(x)))
+        else:
+            return self.nonlinear(self.conv(x))
+
+    def switch_to_deploy(self):
+        kernel, bias = transI_fusebn(self.conv.weight, self.bn)
+        conv = nn.Conv2d(in_channels=self.conv.in_channels, out_channels=self.conv.out_channels, kernel_size=self.conv.kernel_size,
+                                      stride=self.conv.stride, padding=self.conv.padding, dilation=self.conv.dilation, groups=self.conv.groups, bias=True)
+        conv.weight.data = kernel
+        conv.bias.data = bias
+        for para in self.parameters():
+            para.detach_()
+        self.__delattr__('conv')
+        self.__delattr__('bn')
+        self.conv = conv    
+
+class OREPA_3x3_RepConv(nn.Module):
+
+    def __init__(self, in_channels, out_channels, kernel_size,
+                 stride=1, padding=0, dilation=1, groups=1,
+                 internal_channels_1x1_3x3=None,
+                 deploy=False, nonlinear=None, single_init=False):
+        super(OREPA_3x3_RepConv, self).__init__()
+        self.deploy = deploy
+
+        if nonlinear is None:
+            self.nonlinear = nn.Identity()
+        else:
+            self.nonlinear = nonlinear
+
+        self.kernel_size = kernel_size
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.groups = groups
+        assert padding == kernel_size // 2
+
+        self.stride = stride
+        self.padding = padding
+        self.dilation = dilation
+
+        self.branch_counter = 0
+
+        self.weight_rbr_origin = nn.Parameter(torch.Tensor(out_channels, int(in_channels/self.groups), kernel_size, kernel_size))
+        nn.init.kaiming_uniform_(self.weight_rbr_origin, a=math.sqrt(1.0))
+        self.branch_counter += 1
+
+
+        if groups < out_channels:
+            self.weight_rbr_avg_conv = nn.Parameter(torch.Tensor(out_channels, int(in_channels/self.groups), 1, 1))
+            self.weight_rbr_pfir_conv = nn.Parameter(torch.Tensor(out_channels, int(in_channels/self.groups), 1, 1))
+            nn.init.kaiming_uniform_(self.weight_rbr_avg_conv, a=1.0)
+            nn.init.kaiming_uniform_(self.weight_rbr_pfir_conv, a=1.0)
+            self.weight_rbr_avg_conv.data
+            self.weight_rbr_pfir_conv.data
+            self.register_buffer('weight_rbr_avg_avg', torch.ones(kernel_size, kernel_size).mul(1.0/kernel_size/kernel_size))
+            self.branch_counter += 1
+
+        else:
+            raise NotImplementedError
+        self.branch_counter += 1
+
+        if internal_channels_1x1_3x3 is None:
+            internal_channels_1x1_3x3 = in_channels if groups < out_channels else 2 * in_channels   # For mobilenet, it is better to have 2X internal channels
+
+        if internal_channels_1x1_3x3 == in_channels:
+            self.weight_rbr_1x1_kxk_idconv1 = nn.Parameter(torch.zeros(in_channels, int(in_channels/self.groups), 1, 1))
+            id_value = np.zeros((in_channels, int(in_channels/self.groups), 1, 1))
+            for i in range(in_channels):
+                id_value[i, i % int(in_channels/self.groups), 0, 0] = 1
+            id_tensor = torch.from_numpy(id_value).type_as(self.weight_rbr_1x1_kxk_idconv1)
+            self.register_buffer('id_tensor', id_tensor)
+
+        else:
+            self.weight_rbr_1x1_kxk_conv1 = nn.Parameter(torch.Tensor(internal_channels_1x1_3x3, int(in_channels/self.groups), 1, 1))
+            nn.init.kaiming_uniform_(self.weight_rbr_1x1_kxk_conv1, a=math.sqrt(1.0))
+        self.weight_rbr_1x1_kxk_conv2 = nn.Parameter(torch.Tensor(out_channels, int(internal_channels_1x1_3x3/self.groups), kernel_size, kernel_size))
+        nn.init.kaiming_uniform_(self.weight_rbr_1x1_kxk_conv2, a=math.sqrt(1.0))
+        self.branch_counter += 1
+
+        expand_ratio = 8
+        self.weight_rbr_gconv_dw = nn.Parameter(torch.Tensor(in_channels*expand_ratio, 1, kernel_size, kernel_size))
+        self.weight_rbr_gconv_pw = nn.Parameter(torch.Tensor(out_channels, in_channels*expand_ratio, 1, 1))
+        nn.init.kaiming_uniform_(self.weight_rbr_gconv_dw, a=math.sqrt(1.0))
+        nn.init.kaiming_uniform_(self.weight_rbr_gconv_pw, a=math.sqrt(1.0))
+        self.branch_counter += 1
+
+        if out_channels == in_channels and stride == 1:
+            self.branch_counter += 1
+
+        self.vector = nn.Parameter(torch.Tensor(self.branch_counter, self.out_channels))
+        self.bn = nn.BatchNorm2d(out_channels)
+
+        self.fre_init()
+
+        nn.init.constant_(self.vector[0, :], 0.25)    #origin
+        nn.init.constant_(self.vector[1, :], 0.25)      #avg
+        nn.init.constant_(self.vector[2, :], 0.0)      #prior
+        nn.init.constant_(self.vector[3, :], 0.5)    #1x1_kxk
+        nn.init.constant_(self.vector[4, :], 0.5)     #dws_conv
+
+
+    def fre_init(self):
+        prior_tensor = torch.Tensor(self.out_channels, self.kernel_size, self.kernel_size)
+        half_fg = self.out_channels/2
+        for i in range(self.out_channels):
+            for h in range(3):
+                for w in range(3):
+                    if i < half_fg:
+                        prior_tensor[i, h, w] = math.cos(math.pi*(h+0.5)*(i+1)/3)
+                    else:
+                        prior_tensor[i, h, w] = math.cos(math.pi*(w+0.5)*(i+1-half_fg)/3)
+
+        self.register_buffer('weight_rbr_prior', prior_tensor)
+
+    def weight_gen(self):
+
+        weight_rbr_origin = torch.einsum('oihw,o->oihw', self.weight_rbr_origin, self.vector[0, :])
+
+        weight_rbr_avg = torch.einsum('oihw,o->oihw', torch.einsum('oihw,hw->oihw', self.weight_rbr_avg_conv, self.weight_rbr_avg_avg), self.vector[1, :])
+        
+        weight_rbr_pfir = torch.einsum('oihw,o->oihw', torch.einsum('oihw,ohw->oihw', self.weight_rbr_pfir_conv, self.weight_rbr_prior), self.vector[2, :])
+
+        weight_rbr_1x1_kxk_conv1 = None
+        if hasattr(self, 'weight_rbr_1x1_kxk_idconv1'):
+            weight_rbr_1x1_kxk_conv1 = (self.weight_rbr_1x1_kxk_idconv1 + self.id_tensor).squeeze()
+        elif hasattr(self, 'weight_rbr_1x1_kxk_conv1'):
+            weight_rbr_1x1_kxk_conv1 = self.weight_rbr_1x1_kxk_conv1.squeeze()
+        else:
+            raise NotImplementedError
+        weight_rbr_1x1_kxk_conv2 = self.weight_rbr_1x1_kxk_conv2
+
+        if self.groups > 1:
+            g = self.groups
+            t, ig = weight_rbr_1x1_kxk_conv1.size()
+            o, tg, h, w = weight_rbr_1x1_kxk_conv2.size()
+            weight_rbr_1x1_kxk_conv1 = weight_rbr_1x1_kxk_conv1.view(g, int(t/g), ig)
+            weight_rbr_1x1_kxk_conv2 = weight_rbr_1x1_kxk_conv2.view(g, int(o/g), tg, h, w)
+            weight_rbr_1x1_kxk = torch.einsum('gti,gothw->goihw', weight_rbr_1x1_kxk_conv1, weight_rbr_1x1_kxk_conv2).view(o, ig, h, w)
+        else:
+            weight_rbr_1x1_kxk = torch.einsum('ti,othw->oihw', weight_rbr_1x1_kxk_conv1, weight_rbr_1x1_kxk_conv2)
+
+        weight_rbr_1x1_kxk = torch.einsum('oihw,o->oihw', weight_rbr_1x1_kxk, self.vector[3, :])
+
+        weight_rbr_gconv = self.dwsc2full(self.weight_rbr_gconv_dw, self.weight_rbr_gconv_pw, self.in_channels)
+        weight_rbr_gconv = torch.einsum('oihw,o->oihw', weight_rbr_gconv, self.vector[4, :])    
+
+        weight = weight_rbr_origin + weight_rbr_avg + weight_rbr_1x1_kxk + weight_rbr_pfir + weight_rbr_gconv
+
+        return weight
+
+    def dwsc2full(self, weight_dw, weight_pw, groups):
+        
+        t, ig, h, w = weight_dw.size()
+        o, _, _, _ = weight_pw.size()
+        tg = int(t/groups)
+        i = int(ig*groups)
+        weight_dw = weight_dw.view(groups, tg, ig, h, w)
+        weight_pw = weight_pw.squeeze().view(o, groups, tg)
+        
+        weight_dsc = torch.einsum('gtihw,ogt->ogihw', weight_dw, weight_pw)
+        return weight_dsc.view(o, i, h, w)
+
+    def forward(self, inputs):
+        weight = self.weight_gen()
+        out = F.conv2d(inputs, weight, bias=None, stride=self.stride, padding=self.padding, dilation=self.dilation, groups=self.groups)
+
+        return self.nonlinear(self.bn(out))
+
+class RepConv_OREPA(nn.Module):
+
+    def __init__(self, c1, c2, k=3, s=1, padding=1, dilation=1, groups=1, padding_mode='zeros', deploy=False, use_se=False, nonlinear=nn.SiLU()):
+        super(RepConv_OREPA, self).__init__()
+        self.deploy = deploy
+        self.groups = groups
+        self.in_channels = c1
+        self.out_channels = c2
+
+        self.padding = padding
+        self.dilation = dilation
+        self.groups = groups
+
+        assert k == 3
+        assert padding == 1
+
+        padding_11 = padding - k // 2
+
+        if nonlinear is None:
+            self.nonlinearity = nn.Identity()
+        else:
+            self.nonlinearity = nonlinear
+
+        if use_se:
+            self.se = SEBlock(self.out_channels, internal_neurons=self.out_channels // 16)
+        else:
+            self.se = nn.Identity()
+
+        if deploy:
+            self.rbr_reparam = nn.Conv2d(in_channels=self.in_channels, out_channels=self.out_channels, kernel_size=k, stride=s,
+                                      padding=padding, dilation=dilation, groups=groups, bias=True, padding_mode=padding_mode)
+
+        else:
+            self.rbr_identity = nn.BatchNorm2d(num_features=self.in_channels) if self.out_channels == self.in_channels and s == 1 else None
+            self.rbr_dense = OREPA_3x3_RepConv(in_channels=self.in_channels, out_channels=self.out_channels, kernel_size=k, stride=s, padding=padding, groups=groups, dilation=1)
+            self.rbr_1x1 = ConvBN(in_channels=self.in_channels, out_channels=self.out_channels, kernel_size=1, stride=s, padding=padding_11, groups=groups, dilation=1)
+            print('RepVGG Block, identity = ', self.rbr_identity)
+
+
+    def forward(self, inputs):
+        if hasattr(self, 'rbr_reparam'):
+            return self.nonlinearity(self.se(self.rbr_reparam(inputs)))
+
+        if self.rbr_identity is None:
+            id_out = 0
+        else:
+            id_out = self.rbr_identity(inputs)
+
+        out1 = self.rbr_dense(inputs)
+        out2 = self.rbr_1x1(inputs)
+        out3 = id_out
+        out = out1 + out2 + out3
+
+        return self.nonlinearity(self.se(out))
+
+
+    #   Optional. This improves the accuracy and facilitates quantization.
+    #   1.  Cancel the original weight decay on rbr_dense.conv.weight and rbr_1x1.conv.weight.
+    #   2.  Use like this.
+    #       loss = criterion(....)
+    #       for every RepVGGBlock blk:
+    #           loss += weight_decay_coefficient * 0.5 * blk.get_cust_L2()
+    #       optimizer.zero_grad()
+    #       loss.backward()
+
+    # Not used for OREPA
+    def get_custom_L2(self):
+        K3 = self.rbr_dense.weight_gen()
+        K1 = self.rbr_1x1.conv.weight
+        t3 = (self.rbr_dense.bn.weight / ((self.rbr_dense.bn.running_var + self.rbr_dense.bn.eps).sqrt())).reshape(-1, 1, 1, 1).detach()
+        t1 = (self.rbr_1x1.bn.weight / ((self.rbr_1x1.bn.running_var + self.rbr_1x1.bn.eps).sqrt())).reshape(-1, 1, 1, 1).detach()
+
+        l2_loss_circle = (K3 ** 2).sum() - (K3[:, :, 1:2, 1:2] ** 2).sum()      # The L2 loss of the "circle" of weights in 3x3 kernel. Use regular L2 on them.
+        eq_kernel = K3[:, :, 1:2, 1:2] * t3 + K1 * t1                           # The equivalent resultant central point of 3x3 kernel.
+        l2_loss_eq_kernel = (eq_kernel ** 2 / (t3 ** 2 + t1 ** 2)).sum()        # Normalize for an L2 coefficient comparable to regular L2.
+        return l2_loss_eq_kernel + l2_loss_circle
+
+    def get_equivalent_kernel_bias(self):
+        kernel3x3, bias3x3 = self._fuse_bn_tensor(self.rbr_dense)
+        kernel1x1, bias1x1 = self._fuse_bn_tensor(self.rbr_1x1)
+        kernelid, biasid = self._fuse_bn_tensor(self.rbr_identity)
+        return kernel3x3 + self._pad_1x1_to_3x3_tensor(kernel1x1) + kernelid, bias3x3 + bias1x1 + biasid
+
+    def _pad_1x1_to_3x3_tensor(self, kernel1x1):
+        if kernel1x1 is None:
+            return 0
+        else:
+            return torch.nn.functional.pad(kernel1x1, [1,1,1,1])
+
+    def _fuse_bn_tensor(self, branch):
+        if branch is None:
+            return 0, 0
+        if not isinstance(branch, nn.BatchNorm2d):
+            if isinstance(branch, OREPA_3x3_RepConv):
+                kernel = branch.weight_gen()
+            elif isinstance(branch, ConvBN):
+                kernel = branch.conv.weight
+            else:
+                raise NotImplementedError
+            running_mean = branch.bn.running_mean
+            running_var = branch.bn.running_var
+            gamma = branch.bn.weight
+            beta = branch.bn.bias
+            eps = branch.bn.eps
+        else:
+            if not hasattr(self, 'id_tensor'):
+                input_dim = self.in_channels // self.groups
+                kernel_value = np.zeros((self.in_channels, input_dim, 3, 3), dtype=np.float32)
+                for i in range(self.in_channels):
+                    kernel_value[i, i % input_dim, 1, 1] = 1
+                self.id_tensor = torch.from_numpy(kernel_value).to(branch.weight.device)
+            kernel = self.id_tensor
+            running_mean = branch.running_mean
+            running_var = branch.running_var
+            gamma = branch.weight
+            beta = branch.bias
+            eps = branch.eps
+        std = (running_var + eps).sqrt()
+        t = (gamma / std).reshape(-1, 1, 1, 1)
+        return kernel * t, beta - running_mean * gamma / std
+
+    def switch_to_deploy(self):
+        if hasattr(self, 'rbr_reparam'):
+            return
+        print(f"RepConv_OREPA.switch_to_deploy")
+        kernel, bias = self.get_equivalent_kernel_bias()
+        self.rbr_reparam = nn.Conv2d(in_channels=self.rbr_dense.in_channels, out_channels=self.rbr_dense.out_channels,
+                                     kernel_size=self.rbr_dense.kernel_size, stride=self.rbr_dense.stride,
+                                     padding=self.rbr_dense.padding, dilation=self.rbr_dense.dilation, groups=self.rbr_dense.groups, bias=True)
+        self.rbr_reparam.weight.data = kernel
+        self.rbr_reparam.bias.data = bias
+        for para in self.parameters():
+            para.detach_()
+        self.__delattr__('rbr_dense')
+        self.__delattr__('rbr_1x1')
+        if hasattr(self, 'rbr_identity'):
+            self.__delattr__('rbr_identity') 
+
+##### end of orepa #####
+
+
+##### swin transformer #####    
+    
+class WindowAttention(nn.Module):
+
+    def __init__(self, dim, window_size, num_heads, qkv_bias=True, qk_scale=None, attn_drop=0., proj_drop=0.):
+
+        super().__init__()
+        self.dim = dim
+        self.window_size = window_size  # Wh, Ww
+        self.num_heads = num_heads
+        head_dim = dim // num_heads
+        self.scale = qk_scale or head_dim ** -0.5
+
+        # define a parameter table of relative position bias
+        self.relative_position_bias_table = nn.Parameter(
+            torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads))  # 2*Wh-1 * 2*Ww-1, nH
+
+        # get pair-wise relative position index for each token inside the window
+        coords_h = torch.arange(self.window_size[0])
+        coords_w = torch.arange(self.window_size[1])
+        coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Ww
+        coords_flatten = torch.flatten(coords, 1)  # 2, Wh*Ww
+        relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :]  # 2, Wh*Ww, Wh*Ww
+        relative_coords = relative_coords.permute(1, 2, 0).contiguous()  # Wh*Ww, Wh*Ww, 2
+        relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0
+        relative_coords[:, :, 1] += self.window_size[1] - 1
+        relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1
+        relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Ww
+        self.register_buffer("relative_position_index", relative_position_index)
+
+        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
+        self.attn_drop = nn.Dropout(attn_drop)
+        self.proj = nn.Linear(dim, dim)
+        self.proj_drop = nn.Dropout(proj_drop)
+
+        nn.init.normal_(self.relative_position_bias_table, std=.02)
+        self.softmax = nn.Softmax(dim=-1)
+
+    def forward(self, x, mask=None):
+
+        B_, N, C = x.shape
+        qkv = self.qkv(x).reshape(B_, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
+        q, k, v = qkv[0], qkv[1], qkv[2]  # make torchscript happy (cannot use tensor as tuple)
+
+        q = q * self.scale
+        attn = (q @ k.transpose(-2, -1))
+
+        relative_position_bias = self.relative_position_bias_table[self.relative_position_index.view(-1)].view(
+            self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1)  # Wh*Ww,Wh*Ww,nH
+        relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Ww
+        attn = attn + relative_position_bias.unsqueeze(0)
+
+        if mask is not None:
+            nW = mask.shape[0]
+            attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0)
+            attn = attn.view(-1, self.num_heads, N, N)
+            attn = self.softmax(attn)
+        else:
+            attn = self.softmax(attn)
+
+        attn = self.attn_drop(attn)
+
+        # print(attn.dtype, v.dtype)
+        try:
+            x = (attn @ v).transpose(1, 2).reshape(B_, N, C)
+        except:
+            #print(attn.dtype, v.dtype)
+            x = (attn.half() @ v).transpose(1, 2).reshape(B_, N, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x
+
+class Mlp(nn.Module):
+
+    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.SiLU, drop=0.):
+        super().__init__()
+        out_features = out_features or in_features
+        hidden_features = hidden_features or in_features
+        self.fc1 = nn.Linear(in_features, hidden_features)
+        self.act = act_layer()
+        self.fc2 = nn.Linear(hidden_features, out_features)
+        self.drop = nn.Dropout(drop)
+
+    def forward(self, x):
+        x = self.fc1(x)
+        x = self.act(x)
+        x = self.drop(x)
+        x = self.fc2(x)
+        x = self.drop(x)
+        return x
+
+def window_partition(x, window_size):
+
+    B, H, W, C = x.shape
+    assert H % window_size == 0, 'feature map h and w can not divide by window size'
+    x = x.view(B, H // window_size, window_size, W // window_size, window_size, C)
+    windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)
+    return windows
+
+def window_reverse(windows, window_size, H, W):
+    
+    B = int(windows.shape[0] / (H * W / window_size / window_size))
+    x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1)
+    x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)
+    return x
+
+
+class SwinTransformerLayer(nn.Module):
+
+    def __init__(self, dim, num_heads, window_size=8, shift_size=0,
+                 mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., drop_path=0.,
+                 act_layer=nn.SiLU, norm_layer=nn.LayerNorm):
+        super().__init__()
+        self.dim = dim
+        self.num_heads = num_heads
+        self.window_size = window_size
+        self.shift_size = shift_size
+        self.mlp_ratio = mlp_ratio
+        # if min(self.input_resolution) <= self.window_size:
+        #     # if window size is larger than input resolution, we don't partition windows
+        #     self.shift_size = 0
+        #     self.window_size = min(self.input_resolution)
+        assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size"
+
+        self.norm1 = norm_layer(dim)
+        self.attn = WindowAttention(
+            dim, window_size=(self.window_size, self.window_size), num_heads=num_heads,
+            qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop)
+
+        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+        self.norm2 = norm_layer(dim)
+        mlp_hidden_dim = int(dim * mlp_ratio)
+        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)
+
+    def create_mask(self, H, W):
+        # calculate attention mask for SW-MSA
+        img_mask = torch.zeros((1, H, W, 1))  # 1 H W 1
+        h_slices = (slice(0, -self.window_size),
+                    slice(-self.window_size, -self.shift_size),
+                    slice(-self.shift_size, None))
+        w_slices = (slice(0, -self.window_size),
+                    slice(-self.window_size, -self.shift_size),
+                    slice(-self.shift_size, None))
+        cnt = 0
+        for h in h_slices:
+            for w in w_slices:
+                img_mask[:, h, w, :] = cnt
+                cnt += 1
+
+        mask_windows = window_partition(img_mask, self.window_size)  # nW, window_size, window_size, 1
+        mask_windows = mask_windows.view(-1, self.window_size * self.window_size)
+        attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)
+        attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0))
+
+        return attn_mask
+
+    def forward(self, x):
+        # reshape x[b c h w] to x[b l c]
+        _, _, H_, W_ = x.shape
+
+        Padding = False
+        if min(H_, W_) < self.window_size or H_ % self.window_size!=0 or W_ % self.window_size!=0:
+            Padding = True
+            # print(f'img_size {min(H_, W_)} is less than (or not divided by) window_size {self.window_size}, Padding.')
+            pad_r = (self.window_size - W_ % self.window_size) % self.window_size
+            pad_b = (self.window_size - H_ % self.window_size) % self.window_size
+            x = F.pad(x, (0, pad_r, 0, pad_b))
+
+        # print('2', x.shape)
+        B, C, H, W = x.shape
+        L = H * W
+        x = x.permute(0, 2, 3, 1).contiguous().view(B, L, C)  # b, L, c
+
+        # create mask from init to forward
+        if self.shift_size > 0:
+            attn_mask = self.create_mask(H, W).to(x.device)
+        else:
+            attn_mask = None
+
+        shortcut = x
+        x = self.norm1(x)
+        x = x.view(B, H, W, C)
+
+        # cyclic shift
+        if self.shift_size > 0:
+            shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2))
+        else:
+            shifted_x = x
+
+        # partition windows
+        x_windows = window_partition(shifted_x, self.window_size)  # nW*B, window_size, window_size, C
+        x_windows = x_windows.view(-1, self.window_size * self.window_size, C)  # nW*B, window_size*window_size, C
+
+        # W-MSA/SW-MSA
+        attn_windows = self.attn(x_windows, mask=attn_mask)  # nW*B, window_size*window_size, C
+
+        # merge windows
+        attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)
+        shifted_x = window_reverse(attn_windows, self.window_size, H, W)  # B H' W' C
+
+        # reverse cyclic shift
+        if self.shift_size > 0:
+            x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2))
+        else:
+            x = shifted_x
+        x = x.view(B, H * W, C)
+
+        # FFN
+        x = shortcut + self.drop_path(x)
+        x = x + self.drop_path(self.mlp(self.norm2(x)))
+
+        x = x.permute(0, 2, 1).contiguous().view(-1, C, H, W)  # b c h w
+
+        if Padding:
+            x = x[:, :, :H_, :W_]  # reverse padding
+
+        return x
+
+
+class SwinTransformerBlock(nn.Module):
+    def __init__(self, c1, c2, num_heads, num_layers, window_size=8):
+        super().__init__()
+        self.conv = None
+        if c1 != c2:
+            self.conv = Conv(c1, c2)
+
+        # remove input_resolution
+        self.blocks = nn.Sequential(*[SwinTransformerLayer(dim=c2, num_heads=num_heads, window_size=window_size,
+                                 shift_size=0 if (i % 2 == 0) else window_size // 2) for i in range(num_layers)])
+
+    def forward(self, x):
+        if self.conv is not None:
+            x = self.conv(x)
+        x = self.blocks(x)
+        return x
+
+
+class STCSPA(nn.Module):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(STCSPA, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c1, c_, 1, 1)
+        self.cv3 = Conv(2 * c_, c2, 1, 1)
+        num_heads = c_ // 32
+        self.m = SwinTransformerBlock(c_, c_, num_heads, n)
+        #self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        y1 = self.m(self.cv1(x))
+        y2 = self.cv2(x)
+        return self.cv3(torch.cat((y1, y2), dim=1))
+
+
+class STCSPB(nn.Module):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(STCSPB, self).__init__()
+        c_ = int(c2)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_, c_, 1, 1)
+        self.cv3 = Conv(2 * c_, c2, 1, 1)
+        num_heads = c_ // 32
+        self.m = SwinTransformerBlock(c_, c_, num_heads, n)
+        #self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        x1 = self.cv1(x)
+        y1 = self.m(x1)
+        y2 = self.cv2(x1)
+        return self.cv3(torch.cat((y1, y2), dim=1))
+
+
+class STCSPC(nn.Module):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(STCSPC, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c1, c_, 1, 1)
+        self.cv3 = Conv(c_, c_, 1, 1)
+        self.cv4 = Conv(2 * c_, c2, 1, 1)
+        num_heads = c_ // 32
+        self.m = SwinTransformerBlock(c_, c_, num_heads, n)
+        #self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        y1 = self.cv3(self.m(self.cv1(x)))
+        y2 = self.cv2(x)
+        return self.cv4(torch.cat((y1, y2), dim=1))
+
+##### end of swin transformer #####   
+
+
+##### swin transformer v2 ##### 
+  
+class WindowAttention_v2(nn.Module):
+
+    def __init__(self, dim, window_size, num_heads, qkv_bias=True, attn_drop=0., proj_drop=0.,
+                 pretrained_window_size=[0, 0]):
+
+        super().__init__()
+        self.dim = dim
+        self.window_size = window_size  # Wh, Ww
+        self.pretrained_window_size = pretrained_window_size
+        self.num_heads = num_heads
+
+        self.logit_scale = nn.Parameter(torch.log(10 * torch.ones((num_heads, 1, 1))), requires_grad=True)
+
+        # mlp to generate continuous relative position bias
+        self.cpb_mlp = nn.Sequential(nn.Linear(2, 512, bias=True),
+                                     nn.ReLU(inplace=True),
+                                     nn.Linear(512, num_heads, bias=False))
+
+        # get relative_coords_table
+        relative_coords_h = torch.arange(-(self.window_size[0] - 1), self.window_size[0], dtype=torch.float32)
+        relative_coords_w = torch.arange(-(self.window_size[1] - 1), self.window_size[1], dtype=torch.float32)
+        relative_coords_table = torch.stack(
+            torch.meshgrid([relative_coords_h,
+                            relative_coords_w])).permute(1, 2, 0).contiguous().unsqueeze(0)  # 1, 2*Wh-1, 2*Ww-1, 2
+        if pretrained_window_size[0] > 0:
+            relative_coords_table[:, :, :, 0] /= (pretrained_window_size[0] - 1)
+            relative_coords_table[:, :, :, 1] /= (pretrained_window_size[1] - 1)
+        else:
+            relative_coords_table[:, :, :, 0] /= (self.window_size[0] - 1)
+            relative_coords_table[:, :, :, 1] /= (self.window_size[1] - 1)
+        relative_coords_table *= 8  # normalize to -8, 8
+        relative_coords_table = torch.sign(relative_coords_table) * torch.log2(
+            torch.abs(relative_coords_table) + 1.0) / np.log2(8)
+
+        self.register_buffer("relative_coords_table", relative_coords_table)
+
+        # get pair-wise relative position index for each token inside the window
+        coords_h = torch.arange(self.window_size[0])
+        coords_w = torch.arange(self.window_size[1])
+        coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Ww
+        coords_flatten = torch.flatten(coords, 1)  # 2, Wh*Ww
+        relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :]  # 2, Wh*Ww, Wh*Ww
+        relative_coords = relative_coords.permute(1, 2, 0).contiguous()  # Wh*Ww, Wh*Ww, 2
+        relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0
+        relative_coords[:, :, 1] += self.window_size[1] - 1
+        relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1
+        relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Ww
+        self.register_buffer("relative_position_index", relative_position_index)
+
+        self.qkv = nn.Linear(dim, dim * 3, bias=False)
+        if qkv_bias:
+            self.q_bias = nn.Parameter(torch.zeros(dim))
+            self.v_bias = nn.Parameter(torch.zeros(dim))
+        else:
+            self.q_bias = None
+            self.v_bias = None
+        self.attn_drop = nn.Dropout(attn_drop)
+        self.proj = nn.Linear(dim, dim)
+        self.proj_drop = nn.Dropout(proj_drop)
+        self.softmax = nn.Softmax(dim=-1)
+
+    def forward(self, x, mask=None):
+        
+        B_, N, C = x.shape
+        qkv_bias = None
+        if self.q_bias is not None:
+            qkv_bias = torch.cat((self.q_bias, torch.zeros_like(self.v_bias, requires_grad=False), self.v_bias))
+        qkv = F.linear(input=x, weight=self.qkv.weight, bias=qkv_bias)
+        qkv = qkv.reshape(B_, N, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)
+        q, k, v = qkv[0], qkv[1], qkv[2]  # make torchscript happy (cannot use tensor as tuple)
+
+        # cosine attention
+        attn = (F.normalize(q, dim=-1) @ F.normalize(k, dim=-1).transpose(-2, -1))
+        logit_scale = torch.clamp(self.logit_scale, max=torch.log(torch.tensor(1. / 0.01))).exp()
+        attn = attn * logit_scale
+
+        relative_position_bias_table = self.cpb_mlp(self.relative_coords_table).view(-1, self.num_heads)
+        relative_position_bias = relative_position_bias_table[self.relative_position_index.view(-1)].view(
+            self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1)  # Wh*Ww,Wh*Ww,nH
+        relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Ww
+        relative_position_bias = 16 * torch.sigmoid(relative_position_bias)
+        attn = attn + relative_position_bias.unsqueeze(0)
+
+        if mask is not None:
+            nW = mask.shape[0]
+            attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0)
+            attn = attn.view(-1, self.num_heads, N, N)
+            attn = self.softmax(attn)
+        else:
+            attn = self.softmax(attn)
+
+        attn = self.attn_drop(attn)
+
+        try:
+            x = (attn @ v).transpose(1, 2).reshape(B_, N, C)
+        except:
+            x = (attn.half() @ v).transpose(1, 2).reshape(B_, N, C)
+            
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x
+
+    def extra_repr(self) -> str:
+        return f'dim={self.dim}, window_size={self.window_size}, ' \
+               f'pretrained_window_size={self.pretrained_window_size}, num_heads={self.num_heads}'
+
+    def flops(self, N):
+        # calculate flops for 1 window with token length of N
+        flops = 0
+        # qkv = self.qkv(x)
+        flops += N * self.dim * 3 * self.dim
+        # attn = (q @ k.transpose(-2, -1))
+        flops += self.num_heads * N * (self.dim // self.num_heads) * N
+        #  x = (attn @ v)
+        flops += self.num_heads * N * N * (self.dim // self.num_heads)
+        # x = self.proj(x)
+        flops += N * self.dim * self.dim
+        return flops
+    
+class Mlp_v2(nn.Module):
+    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.SiLU, drop=0.):
+        super().__init__()
+        out_features = out_features or in_features
+        hidden_features = hidden_features or in_features
+        self.fc1 = nn.Linear(in_features, hidden_features)
+        self.act = act_layer()
+        self.fc2 = nn.Linear(hidden_features, out_features)
+        self.drop = nn.Dropout(drop)
+
+    def forward(self, x):
+        x = self.fc1(x)
+        x = self.act(x)
+        x = self.drop(x)
+        x = self.fc2(x)
+        x = self.drop(x)
+        return x
+
+
+def window_partition_v2(x, window_size):
+    
+    B, H, W, C = x.shape
+    x = x.view(B, H // window_size, window_size, W // window_size, window_size, C)
+    windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)
+    return windows
+
+
+def window_reverse_v2(windows, window_size, H, W):
+    
+    B = int(windows.shape[0] / (H * W / window_size / window_size))
+    x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1)
+    x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)
+    return x
+
+
+class SwinTransformerLayer_v2(nn.Module):
+
+    def __init__(self, dim, num_heads, window_size=7, shift_size=0,
+                 mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0.,
+                 act_layer=nn.SiLU, norm_layer=nn.LayerNorm, pretrained_window_size=0):
+        super().__init__()
+        self.dim = dim
+        #self.input_resolution = input_resolution
+        self.num_heads = num_heads
+        self.window_size = window_size
+        self.shift_size = shift_size
+        self.mlp_ratio = mlp_ratio
+        #if min(self.input_resolution) <= self.window_size:
+        #    # if window size is larger than input resolution, we don't partition windows
+        #    self.shift_size = 0
+        #    self.window_size = min(self.input_resolution)
+        assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size"
+
+        self.norm1 = norm_layer(dim)
+        self.attn = WindowAttention_v2(
+            dim, window_size=(self.window_size, self.window_size), num_heads=num_heads,
+            qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop,
+            pretrained_window_size=(pretrained_window_size, pretrained_window_size))
+
+        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+        self.norm2 = norm_layer(dim)
+        mlp_hidden_dim = int(dim * mlp_ratio)
+        self.mlp = Mlp_v2(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)
+
+    def create_mask(self, H, W):
+        # calculate attention mask for SW-MSA
+        img_mask = torch.zeros((1, H, W, 1))  # 1 H W 1
+        h_slices = (slice(0, -self.window_size),
+                    slice(-self.window_size, -self.shift_size),
+                    slice(-self.shift_size, None))
+        w_slices = (slice(0, -self.window_size),
+                    slice(-self.window_size, -self.shift_size),
+                    slice(-self.shift_size, None))
+        cnt = 0
+        for h in h_slices:
+            for w in w_slices:
+                img_mask[:, h, w, :] = cnt
+                cnt += 1
+
+        mask_windows = window_partition(img_mask, self.window_size)  # nW, window_size, window_size, 1
+        mask_windows = mask_windows.view(-1, self.window_size * self.window_size)
+        attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)
+        attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0))
+
+        return attn_mask
+
+    def forward(self, x):
+        # reshape x[b c h w] to x[b l c]
+        _, _, H_, W_ = x.shape
+
+        Padding = False
+        if min(H_, W_) < self.window_size or H_ % self.window_size!=0 or W_ % self.window_size!=0:
+            Padding = True
+            # print(f'img_size {min(H_, W_)} is less than (or not divided by) window_size {self.window_size}, Padding.')
+            pad_r = (self.window_size - W_ % self.window_size) % self.window_size
+            pad_b = (self.window_size - H_ % self.window_size) % self.window_size
+            x = F.pad(x, (0, pad_r, 0, pad_b))
+
+        # print('2', x.shape)
+        B, C, H, W = x.shape
+        L = H * W
+        x = x.permute(0, 2, 3, 1).contiguous().view(B, L, C)  # b, L, c
+
+        # create mask from init to forward
+        if self.shift_size > 0:
+            attn_mask = self.create_mask(H, W).to(x.device)
+        else:
+            attn_mask = None
+
+        shortcut = x
+        x = x.view(B, H, W, C)
+
+        # cyclic shift
+        if self.shift_size > 0:
+            shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2))
+        else:
+            shifted_x = x
+
+        # partition windows
+        x_windows = window_partition_v2(shifted_x, self.window_size)  # nW*B, window_size, window_size, C
+        x_windows = x_windows.view(-1, self.window_size * self.window_size, C)  # nW*B, window_size*window_size, C
+
+        # W-MSA/SW-MSA
+        attn_windows = self.attn(x_windows, mask=attn_mask)  # nW*B, window_size*window_size, C
+
+        # merge windows
+        attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)
+        shifted_x = window_reverse_v2(attn_windows, self.window_size, H, W)  # B H' W' C
+
+        # reverse cyclic shift
+        if self.shift_size > 0:
+            x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2))
+        else:
+            x = shifted_x
+        x = x.view(B, H * W, C)
+        x = shortcut + self.drop_path(self.norm1(x))
+
+        # FFN
+        x = x + self.drop_path(self.norm2(self.mlp(x)))
+        x = x.permute(0, 2, 1).contiguous().view(-1, C, H, W)  # b c h w
+        
+        if Padding:
+            x = x[:, :, :H_, :W_]  # reverse padding
+
+        return x
+
+    def extra_repr(self) -> str:
+        return f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, " \
+               f"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}"
+
+    def flops(self):
+        flops = 0
+        H, W = self.input_resolution
+        # norm1
+        flops += self.dim * H * W
+        # W-MSA/SW-MSA
+        nW = H * W / self.window_size / self.window_size
+        flops += nW * self.attn.flops(self.window_size * self.window_size)
+        # mlp
+        flops += 2 * H * W * self.dim * self.dim * self.mlp_ratio
+        # norm2
+        flops += self.dim * H * W
+        return flops
+
+
+class SwinTransformer2Block(nn.Module):
+    def __init__(self, c1, c2, num_heads, num_layers, window_size=7):
+        super().__init__()
+        self.conv = None
+        if c1 != c2:
+            self.conv = Conv(c1, c2)
+
+        # remove input_resolution
+        self.blocks = nn.Sequential(*[SwinTransformerLayer_v2(dim=c2, num_heads=num_heads, window_size=window_size,
+                                 shift_size=0 if (i % 2 == 0) else window_size // 2) for i in range(num_layers)])
+
+    def forward(self, x):
+        if self.conv is not None:
+            x = self.conv(x)
+        x = self.blocks(x)
+        return x
+
+
+class ST2CSPA(nn.Module):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(ST2CSPA, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c1, c_, 1, 1)
+        self.cv3 = Conv(2 * c_, c2, 1, 1)
+        num_heads = c_ // 32
+        self.m = SwinTransformer2Block(c_, c_, num_heads, n)
+        #self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        y1 = self.m(self.cv1(x))
+        y2 = self.cv2(x)
+        return self.cv3(torch.cat((y1, y2), dim=1))
+
+
+class ST2CSPB(nn.Module):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(ST2CSPB, self).__init__()
+        c_ = int(c2)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c_, c_, 1, 1)
+        self.cv3 = Conv(2 * c_, c2, 1, 1)
+        num_heads = c_ // 32
+        self.m = SwinTransformer2Block(c_, c_, num_heads, n)
+        #self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        x1 = self.cv1(x)
+        y1 = self.m(x1)
+        y2 = self.cv2(x1)
+        return self.cv3(torch.cat((y1, y2), dim=1))
+
+
+class ST2CSPC(nn.Module):
+    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
+        super(ST2CSPC, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, 1, 1)
+        self.cv2 = Conv(c1, c_, 1, 1)
+        self.cv3 = Conv(c_, c_, 1, 1)
+        self.cv4 = Conv(2 * c_, c2, 1, 1)
+        num_heads = c_ // 32
+        self.m = SwinTransformer2Block(c_, c_, num_heads, n)
+        #self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])
+
+    def forward(self, x):
+        y1 = self.cv3(self.m(self.cv1(x)))
+        y2 = self.cv2(x)
+        return self.cv4(torch.cat((y1, y2), dim=1))
+
+##### end of swin transformer v2 #####   

models/models/experimental.py

@@ -0,0 +1,272 @@
+import numpy as np
+import random
+import torch
+import torch.nn as nn
+
+from models.common import Conv, DWConv
+from utils.google_utils import attempt_download
+
+
+class CrossConv(nn.Module):
+    # Cross Convolution Downsample
+    def __init__(self, c1, c2, k=3, s=1, g=1, e=1.0, shortcut=False):
+        # ch_in, ch_out, kernel, stride, groups, expansion, shortcut
+        super(CrossConv, self).__init__()
+        c_ = int(c2 * e)  # hidden channels
+        self.cv1 = Conv(c1, c_, (1, k), (1, s))
+        self.cv2 = Conv(c_, c2, (k, 1), (s, 1), g=g)
+        self.add = shortcut and c1 == c2
+
+    def forward(self, x):
+        return x + self.cv2(self.cv1(x)) if self.add else self.cv2(self.cv1(x))
+
+
+class Sum(nn.Module):
+    # Weighted sum of 2 or more layers https://arxiv.org/abs/1911.09070
+    def __init__(self, n, weight=False):  # n: number of inputs
+        super(Sum, self).__init__()
+        self.weight = weight  # apply weights boolean
+        self.iter = range(n - 1)  # iter object
+        if weight:
+            self.w = nn.Parameter(-torch.arange(1., n) / 2, requires_grad=True)  # layer weights
+
+    def forward(self, x):
+        y = x[0]  # no weight
+        if self.weight:
+            w = torch.sigmoid(self.w) * 2
+            for i in self.iter:
+                y = y + x[i + 1] * w[i]
+        else:
+            for i in self.iter:
+                y = y + x[i + 1]
+        return y
+
+
+class MixConv2d(nn.Module):
+    # Mixed Depthwise Conv https://arxiv.org/abs/1907.09595
+    def __init__(self, c1, c2, k=(1, 3), s=1, equal_ch=True):
+        super(MixConv2d, self).__init__()
+        groups = len(k)
+        if equal_ch:  # equal c_ per group
+            i = torch.linspace(0, groups - 1E-6, c2).floor()  # c2 indices
+            c_ = [(i == g).sum() for g in range(groups)]  # intermediate channels
+        else:  # equal weight.numel() per group
+            b = [c2] + [0] * groups
+            a = np.eye(groups + 1, groups, k=-1)
+            a -= np.roll(a, 1, axis=1)
+            a *= np.array(k) ** 2
+            a[0] = 1
+            c_ = np.linalg.lstsq(a, b, rcond=None)[0].round()  # solve for equal weight indices, ax = b
+
+        self.m = nn.ModuleList([nn.Conv2d(c1, int(c_[g]), k[g], s, k[g] // 2, bias=False) for g in range(groups)])
+        self.bn = nn.BatchNorm2d(c2)
+        self.act = nn.LeakyReLU(0.1, inplace=True)
+
+    def forward(self, x):
+        return x + self.act(self.bn(torch.cat([m(x) for m in self.m], 1)))
+
+
+class Ensemble(nn.ModuleList):
+    # Ensemble of models
+    def __init__(self):
+        super(Ensemble, self).__init__()
+
+    def forward(self, x, augment=False):
+        y = []
+        for module in self:
+            y.append(module(x, augment)[0])
+        # y = torch.stack(y).max(0)[0]  # max ensemble
+        # y = torch.stack(y).mean(0)  # mean ensemble
+        y = torch.cat(y, 1)  # nms ensemble
+        return y, None  # inference, train output
+
+
+
+
+
+class ORT_NMS(torch.autograd.Function):
+    '''ONNX-Runtime NMS operation'''
+    @staticmethod
+    def forward(ctx,
+                boxes,
+                scores,
+                max_output_boxes_per_class=torch.tensor([100]),
+                iou_threshold=torch.tensor([0.45]),
+                score_threshold=torch.tensor([0.25])):
+        device = boxes.device
+        batch = scores.shape[0]
+        num_det = random.randint(0, 100)
+        batches = torch.randint(0, batch, (num_det,)).sort()[0].to(device)
+        idxs = torch.arange(100, 100 + num_det).to(device)
+        zeros = torch.zeros((num_det,), dtype=torch.int64).to(device)
+        selected_indices = torch.cat([batches[None], zeros[None], idxs[None]], 0).T.contiguous()
+        selected_indices = selected_indices.to(torch.int64)
+        return selected_indices
+
+    @staticmethod
+    def symbolic(g, boxes, scores, max_output_boxes_per_class, iou_threshold, score_threshold):
+        return g.op("NonMaxSuppression", boxes, scores, max_output_boxes_per_class, iou_threshold, score_threshold)
+
+
+class TRT_NMS(torch.autograd.Function):
+    '''TensorRT NMS operation'''
+    @staticmethod
+    def forward(
+        ctx,
+        boxes,
+        scores,
+        background_class=-1,
+        box_coding=1,
+        iou_threshold=0.45,
+        max_output_boxes=100,
+        plugin_version="1",
+        score_activation=0,
+        score_threshold=0.25,
+    ):
+        batch_size, num_boxes, num_classes = scores.shape
+        num_det = torch.randint(0, max_output_boxes, (batch_size, 1), dtype=torch.int32)
+        det_boxes = torch.randn(batch_size, max_output_boxes, 4)
+        det_scores = torch.randn(batch_size, max_output_boxes)
+        det_classes = torch.randint(0, num_classes, (batch_size, max_output_boxes), dtype=torch.int32)
+        return num_det, det_boxes, det_scores, det_classes
+
+    @staticmethod
+    def symbolic(g,
+                 boxes,
+                 scores,
+                 background_class=-1,
+                 box_coding=1,
+                 iou_threshold=0.45,
+                 max_output_boxes=100,
+                 plugin_version="1",
+                 score_activation=0,
+                 score_threshold=0.25):
+        out = g.op("TRT::EfficientNMS_TRT",
+                   boxes,
+                   scores,
+                   background_class_i=background_class,
+                   box_coding_i=box_coding,
+                   iou_threshold_f=iou_threshold,
+                   max_output_boxes_i=max_output_boxes,
+                   plugin_version_s=plugin_version,
+                   score_activation_i=score_activation,
+                   score_threshold_f=score_threshold,
+                   outputs=4)
+        nums, boxes, scores, classes = out
+        return nums, boxes, scores, classes
+
+
+class ONNX_ORT(nn.Module):
+    '''onnx module with ONNX-Runtime NMS operation.'''
+    def __init__(self, max_obj=100, iou_thres=0.45, score_thres=0.25, max_wh=640, device=None, n_classes=80):
+        super().__init__()
+        self.device = device if device else torch.device("cpu")
+        self.max_obj = torch.tensor([max_obj]).to(device)
+        self.iou_threshold = torch.tensor([iou_thres]).to(device)
+        self.score_threshold = torch.tensor([score_thres]).to(device)
+        self.max_wh = max_wh # if max_wh != 0 : non-agnostic else : agnostic
+        self.convert_matrix = torch.tensor([[1, 0, 1, 0], [0, 1, 0, 1], [-0.5, 0, 0.5, 0], [0, -0.5, 0, 0.5]],
+                                           dtype=torch.float32,
+                                           device=self.device)
+        self.n_classes=n_classes
+
+    def forward(self, x):
+        boxes = x[:, :, :4]
+        conf = x[:, :, 4:5]
+        scores = x[:, :, 5:]
+        if self.n_classes == 1:
+            scores = conf # for models with one class, cls_loss is 0 and cls_conf is always 0.5,
+                                 # so there is no need to multiplicate.
+        else:
+            scores *= conf  # conf = obj_conf * cls_conf
+        boxes @= self.convert_matrix
+        max_score, category_id = scores.max(2, keepdim=True)
+        dis = category_id.float() * self.max_wh
+        nmsbox = boxes + dis
+        max_score_tp = max_score.transpose(1, 2).contiguous()
+        selected_indices = ORT_NMS.apply(nmsbox, max_score_tp, self.max_obj, self.iou_threshold, self.score_threshold)
+        X, Y = selected_indices[:, 0], selected_indices[:, 2]
+        selected_boxes = boxes[X, Y, :]
+        selected_categories = category_id[X, Y, :].float()
+        selected_scores = max_score[X, Y, :]
+        X = X.unsqueeze(1).float()
+        return torch.cat([X, selected_boxes, selected_categories, selected_scores], 1)
+
+class ONNX_TRT(nn.Module):
+    '''onnx module with TensorRT NMS operation.'''
+    def __init__(self, max_obj=100, iou_thres=0.45, score_thres=0.25, max_wh=None ,device=None, n_classes=80):
+        super().__init__()
+        assert max_wh is None
+        self.device = device if device else torch.device('cpu')
+        self.background_class = -1,
+        self.box_coding = 1,
+        self.iou_threshold = iou_thres
+        self.max_obj = max_obj
+        self.plugin_version = '1'
+        self.score_activation = 0
+        self.score_threshold = score_thres
+        self.n_classes=n_classes
+
+    def forward(self, x):
+        boxes = x[:, :, :4]
+        conf = x[:, :, 4:5]
+        scores = x[:, :, 5:]
+        if self.n_classes == 1:
+            scores = conf # for models with one class, cls_loss is 0 and cls_conf is always 0.5,
+                                 # so there is no need to multiplicate.
+        else:
+            scores *= conf  # conf = obj_conf * cls_conf
+        num_det, det_boxes, det_scores, det_classes = TRT_NMS.apply(boxes, scores, self.background_class, self.box_coding,
+                                                                    self.iou_threshold, self.max_obj,
+                                                                    self.plugin_version, self.score_activation,
+                                                                    self.score_threshold)
+        return num_det, det_boxes, det_scores, det_classes
+
+
+class End2End(nn.Module):
+    '''export onnx or tensorrt model with NMS operation.'''
+    def __init__(self, model, max_obj=100, iou_thres=0.45, score_thres=0.25, max_wh=None, device=None, n_classes=80):
+        super().__init__()
+        device = device if device else torch.device('cpu')
+        assert isinstance(max_wh,(int)) or max_wh is None
+        self.model = model.to(device)
+        self.model.model[-1].end2end = True
+        self.patch_model = ONNX_TRT if max_wh is None else ONNX_ORT
+        self.end2end = self.patch_model(max_obj, iou_thres, score_thres, max_wh, device, n_classes)
+        self.end2end.eval()
+
+    def forward(self, x):
+        x = self.model(x)
+        x = self.end2end(x)
+        return x
+
+
+
+
+
+def attempt_load(weights, map_location=None):
+    # Loads an ensemble of models weights=[a,b,c] or a single model weights=[a] or weights=a
+    model = Ensemble()
+    for w in weights if isinstance(weights, list) else [weights]:
+        attempt_download(w)
+        ckpt = torch.load(w, map_location=map_location)  # load
+        model.append(ckpt['ema' if ckpt.get('ema') else 'model'].float().fuse().eval())  # FP32 model
+    
+    # Compatibility updates
+    for m in model.modules():
+        if type(m) in [nn.Hardswish, nn.LeakyReLU, nn.ReLU, nn.ReLU6, nn.SiLU]:
+            m.inplace = True  # pytorch 1.7.0 compatibility
+        elif type(m) is nn.Upsample:
+            m.recompute_scale_factor = None  # torch 1.11.0 compatibility
+        elif type(m) is Conv:
+            m._non_persistent_buffers_set = set()  # pytorch 1.6.0 compatibility
+    
+    if len(model) == 1:
+        return model[-1]  # return model
+    else:
+        print('Ensemble created with %s\n' % weights)
+        for k in ['names', 'stride']:
+            setattr(model, k, getattr(model[-1], k))
+        return model  # return ensemble
+
+

models/models/yolo.py

@@ -0,0 +1,843 @@
+import argparse
+import logging
+import sys
+from copy import deepcopy
+
+sys.path.append('./')  # to run '$ python *.py' files in subdirectories
+logger = logging.getLogger(__name__)
+import torch
+from models.common import *
+from models.experimental import *
+from utils.autoanchor import check_anchor_order
+from utils.general import make_divisible, check_file, set_logging
+from utils.torch_utils import time_synchronized, fuse_conv_and_bn, model_info, scale_img, initialize_weights, \
+    select_device, copy_attr
+from utils.loss import SigmoidBin
+
+try:
+    import thop  # for FLOPS computation
+except ImportError:
+    thop = None
+
+
+class Detect(nn.Module):
+    stride = None  # strides computed during build
+    export = False  # onnx export
+    end2end = False
+    include_nms = False
+    concat = False
+
+    def __init__(self, nc=80, anchors=(), ch=()):  # detection layer
+        super(Detect, self).__init__()
+        self.nc = nc  # number of classes
+        self.no = nc + 5  # number of outputs per anchor
+        self.nl = len(anchors)  # number of detection layers
+        self.na = len(anchors[0]) // 2  # number of anchors
+        self.grid = [torch.zeros(1)] * self.nl  # init grid
+        a = torch.tensor(anchors).float().view(self.nl, -1, 2)
+        self.register_buffer('anchors', a)  # shape(nl,na,2)
+        self.register_buffer('anchor_grid', a.clone().view(self.nl, 1, -1, 1, 1, 2))  # shape(nl,1,na,1,1,2)
+        self.m = nn.ModuleList(nn.Conv2d(x, self.no * self.na, 1) for x in ch)  # output conv
+
+    def forward(self, x):
+        # x = x.copy()  # for profiling
+        z = []  # inference output
+        self.training |= self.export
+        for i in range(self.nl):
+            x[i] = self.m[i](x[i])  # conv
+            bs, _, ny, nx = x[i].shape  # x(bs,255,20,20) to x(bs,3,20,20,85)
+            x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+
+            if not self.training:  # inference
+                if self.grid[i].shape[2:4] != x[i].shape[2:4]:
+                    self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
+                y = x[i].sigmoid()
+                if not torch.onnx.is_in_onnx_export():
+                    y[..., 0:2] = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                    y[..., 2:4] = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i]  # wh
+                else:
+                    xy, wh, conf = y.split((2, 2, self.nc + 1), 4)  # y.tensor_split((2, 4, 5), 4)  # torch 1.8.0
+                    xy = xy * (2. * self.stride[i]) + (self.stride[i] * (self.grid[i] - 0.5))  # new xy
+                    wh = wh ** 2 * (4 * self.anchor_grid[i].data)  # new wh
+                    y = torch.cat((xy, wh, conf), 4)
+                z.append(y.view(bs, -1, self.no))
+
+        if self.training:
+            out = x
+        elif self.end2end:
+            out = torch.cat(z, 1)
+        elif self.include_nms:
+            z = self.convert(z)
+            out = (z, )
+        elif self.concat:
+            out = torch.cat(z, 1)
+        else:
+            out = (torch.cat(z, 1), x)
+
+        return out
+
+    @staticmethod
+    def _make_grid(nx=20, ny=20):
+        yv, xv = torch.meshgrid([torch.arange(ny), torch.arange(nx)])
+        return torch.stack((xv, yv), 2).view((1, 1, ny, nx, 2)).float()
+
+    def convert(self, z):
+        z = torch.cat(z, 1)
+        box = z[:, :, :4]
+        conf = z[:, :, 4:5]
+        score = z[:, :, 5:]
+        score *= conf
+        convert_matrix = torch.tensor([[1, 0, 1, 0], [0, 1, 0, 1], [-0.5, 0, 0.5, 0], [0, -0.5, 0, 0.5]],
+                                           dtype=torch.float32,
+                                           device=z.device)
+        box @= convert_matrix                          
+        return (box, score)
+
+
+class IDetect(nn.Module):
+    stride = None  # strides computed during build
+    export = False  # onnx export
+    end2end = False
+    include_nms = False
+    concat = False
+
+    def __init__(self, nc=80, anchors=(), ch=()):  # detection layer
+        super(IDetect, self).__init__()
+        self.nc = nc  # number of classes
+        self.no = nc + 5  # number of outputs per anchor
+        self.nl = len(anchors)  # number of detection layers
+        self.na = len(anchors[0]) // 2  # number of anchors
+        self.grid = [torch.zeros(1)] * self.nl  # init grid
+        a = torch.tensor(anchors).float().view(self.nl, -1, 2)
+        self.register_buffer('anchors', a)  # shape(nl,na,2)
+        self.register_buffer('anchor_grid', a.clone().view(self.nl, 1, -1, 1, 1, 2))  # shape(nl,1,na,1,1,2)
+        self.m = nn.ModuleList(nn.Conv2d(x, self.no * self.na, 1) for x in ch)  # output conv
+        
+        self.ia = nn.ModuleList(ImplicitA(x) for x in ch)
+        self.im = nn.ModuleList(ImplicitM(self.no * self.na) for _ in ch)
+
+    def forward(self, x):
+        # x = x.copy()  # for profiling
+        z = []  # inference output
+        self.training |= self.export
+        for i in range(self.nl):
+            x[i] = self.m[i](self.ia[i](x[i]))  # conv
+            x[i] = self.im[i](x[i])
+            bs, _, ny, nx = x[i].shape  # x(bs,255,20,20) to x(bs,3,20,20,85)
+            x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+
+            if not self.training:  # inference
+                if self.grid[i].shape[2:4] != x[i].shape[2:4]:
+                    self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
+
+                y = x[i].sigmoid()
+                y[..., 0:2] = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                y[..., 2:4] = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i]  # wh
+                z.append(y.view(bs, -1, self.no))
+
+        return x if self.training else (torch.cat(z, 1), x)
+    
+    def fuseforward(self, x):
+        # x = x.copy()  # for profiling
+        z = []  # inference output
+        self.training |= self.export
+        for i in range(self.nl):
+            x[i] = self.m[i](x[i])  # conv
+            bs, _, ny, nx = x[i].shape  # x(bs,255,20,20) to x(bs,3,20,20,85)
+            x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+
+            if not self.training:  # inference
+                if self.grid[i].shape[2:4] != x[i].shape[2:4]:
+                    self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
+
+                y = x[i].sigmoid()
+                if not torch.onnx.is_in_onnx_export():
+                    y[..., 0:2] = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                    y[..., 2:4] = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i]  # wh
+                else:
+                    xy, wh, conf = y.split((2, 2, self.nc + 1), 4)  # y.tensor_split((2, 4, 5), 4)  # torch 1.8.0
+                    xy = xy * (2. * self.stride[i]) + (self.stride[i] * (self.grid[i] - 0.5))  # new xy
+                    wh = wh ** 2 * (4 * self.anchor_grid[i].data)  # new wh
+                    y = torch.cat((xy, wh, conf), 4)
+                z.append(y.view(bs, -1, self.no))
+
+        if self.training:
+            out = x
+        elif self.end2end:
+            out = torch.cat(z, 1)
+        elif self.include_nms:
+            z = self.convert(z)
+            out = (z, )
+        elif self.concat:
+            out = torch.cat(z, 1)            
+        else:
+            out = (torch.cat(z, 1), x)
+
+        return out
+    
+    def fuse(self):
+        print("IDetect.fuse")
+        # fuse ImplicitA and Convolution
+        for i in range(len(self.m)):
+            c1,c2,_,_ = self.m[i].weight.shape
+            c1_,c2_, _,_ = self.ia[i].implicit.shape
+            self.m[i].bias += torch.matmul(self.m[i].weight.reshape(c1,c2),self.ia[i].implicit.reshape(c2_,c1_)).squeeze(1)
+
+        # fuse ImplicitM and Convolution
+        for i in range(len(self.m)):
+            c1,c2, _,_ = self.im[i].implicit.shape
+            self.m[i].bias *= self.im[i].implicit.reshape(c2)
+            self.m[i].weight *= self.im[i].implicit.transpose(0,1)
+            
+    @staticmethod
+    def _make_grid(nx=20, ny=20):
+        yv, xv = torch.meshgrid([torch.arange(ny), torch.arange(nx)])
+        return torch.stack((xv, yv), 2).view((1, 1, ny, nx, 2)).float()
+
+    def convert(self, z):
+        z = torch.cat(z, 1)
+        box = z[:, :, :4]
+        conf = z[:, :, 4:5]
+        score = z[:, :, 5:]
+        score *= conf
+        convert_matrix = torch.tensor([[1, 0, 1, 0], [0, 1, 0, 1], [-0.5, 0, 0.5, 0], [0, -0.5, 0, 0.5]],
+                                           dtype=torch.float32,
+                                           device=z.device)
+        box @= convert_matrix                          
+        return (box, score)
+
+
+class IKeypoint(nn.Module):
+    stride = None  # strides computed during build
+    export = False  # onnx export
+
+    def __init__(self, nc=80, anchors=(), nkpt=17, ch=(), inplace=True, dw_conv_kpt=False):  # detection layer
+        super(IKeypoint, self).__init__()
+        self.nc = nc  # number of classes
+        self.nkpt = nkpt
+        self.dw_conv_kpt = dw_conv_kpt
+        self.no_det=(nc + 5)  # number of outputs per anchor for box and class
+        self.no_kpt = 3*self.nkpt ## number of outputs per anchor for keypoints
+        self.no = self.no_det+self.no_kpt
+        self.nl = len(anchors)  # number of detection layers
+        self.na = len(anchors[0]) // 2  # number of anchors
+        self.grid = [torch.zeros(1)] * self.nl  # init grid
+        self.flip_test = False
+        a = torch.tensor(anchors).float().view(self.nl, -1, 2)
+        self.register_buffer('anchors', a)  # shape(nl,na,2)
+        self.register_buffer('anchor_grid', a.clone().view(self.nl, 1, -1, 1, 1, 2))  # shape(nl,1,na,1,1,2)
+        self.m = nn.ModuleList(nn.Conv2d(x, self.no_det * self.na, 1) for x in ch)  # output conv
+        
+        self.ia = nn.ModuleList(ImplicitA(x) for x in ch)
+        self.im = nn.ModuleList(ImplicitM(self.no_det * self.na) for _ in ch)
+        
+        if self.nkpt is not None:
+            if self.dw_conv_kpt: #keypoint head is slightly more complex
+                self.m_kpt = nn.ModuleList(
+                            nn.Sequential(DWConv(x, x, k=3), Conv(x,x),
+                                          DWConv(x, x, k=3), Conv(x, x),
+                                          DWConv(x, x, k=3), Conv(x,x),
+                                          DWConv(x, x, k=3), Conv(x, x),
+                                          DWConv(x, x, k=3), Conv(x, x),
+                                          DWConv(x, x, k=3), nn.Conv2d(x, self.no_kpt * self.na, 1)) for x in ch)
+            else: #keypoint head is a single convolution
+                self.m_kpt = nn.ModuleList(nn.Conv2d(x, self.no_kpt * self.na, 1) for x in ch)
+
+        self.inplace = inplace  # use in-place ops (e.g. slice assignment)
+
+    def forward(self, x):
+        # x = x.copy()  # for profiling
+        z = []  # inference output
+        self.training |= self.export
+        for i in range(self.nl):
+            if self.nkpt is None or self.nkpt==0:
+                x[i] = self.im[i](self.m[i](self.ia[i](x[i])))  # conv
+            else :
+                x[i] = torch.cat((self.im[i](self.m[i](self.ia[i](x[i]))), self.m_kpt[i](x[i])), axis=1)
+
+            bs, _, ny, nx = x[i].shape  # x(bs,255,20,20) to x(bs,3,20,20,85)
+            x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+            x_det = x[i][..., :6]
+            x_kpt = x[i][..., 6:]
+
+            if not self.training:  # inference
+                if self.grid[i].shape[2:4] != x[i].shape[2:4]:
+                    self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
+                kpt_grid_x = self.grid[i][..., 0:1]
+                kpt_grid_y = self.grid[i][..., 1:2]
+
+                if self.nkpt == 0:
+                    y = x[i].sigmoid()
+                else:
+                    y = x_det.sigmoid()
+
+                if self.inplace:
+                    xy = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                    wh = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i].view(1, self.na, 1, 1, 2) # wh
+                    if self.nkpt != 0:
+                        x_kpt[..., 0::3] = (x_kpt[..., ::3] * 2. - 0.5 + kpt_grid_x.repeat(1,1,1,1,17)) * self.stride[i]  # xy
+                        x_kpt[..., 1::3] = (x_kpt[..., 1::3] * 2. - 0.5 + kpt_grid_y.repeat(1,1,1,1,17)) * self.stride[i]  # xy
+                        #x_kpt[..., 0::3] = (x_kpt[..., ::3] + kpt_grid_x.repeat(1,1,1,1,17)) * self.stride[i]  # xy
+                        #x_kpt[..., 1::3] = (x_kpt[..., 1::3] + kpt_grid_y.repeat(1,1,1,1,17)) * self.stride[i]  # xy
+                        #print('=============')
+                        #print(self.anchor_grid[i].shape)
+                        #print(self.anchor_grid[i][...,0].unsqueeze(4).shape)
+                        #print(x_kpt[..., 0::3].shape)
+                        #x_kpt[..., 0::3] = ((x_kpt[..., 0::3].tanh() * 2.) ** 3 * self.anchor_grid[i][...,0].unsqueeze(4).repeat(1,1,1,1,self.nkpt)) + kpt_grid_x.repeat(1,1,1,1,17) * self.stride[i]  # xy
+                        #x_kpt[..., 1::3] = ((x_kpt[..., 1::3].tanh() * 2.) ** 3 * self.anchor_grid[i][...,1].unsqueeze(4).repeat(1,1,1,1,self.nkpt)) + kpt_grid_y.repeat(1,1,1,1,17) * self.stride[i]  # xy
+                        #x_kpt[..., 0::3] = (((x_kpt[..., 0::3].sigmoid() * 4.) ** 2 - 8.) * self.anchor_grid[i][...,0].unsqueeze(4).repeat(1,1,1,1,self.nkpt)) + kpt_grid_x.repeat(1,1,1,1,17) * self.stride[i]  # xy
+                        #x_kpt[..., 1::3] = (((x_kpt[..., 1::3].sigmoid() * 4.) ** 2 - 8.) * self.anchor_grid[i][...,1].unsqueeze(4).repeat(1,1,1,1,self.nkpt)) + kpt_grid_y.repeat(1,1,1,1,17) * self.stride[i]  # xy
+                        x_kpt[..., 2::3] = x_kpt[..., 2::3].sigmoid()
+
+                    y = torch.cat((xy, wh, y[..., 4:], x_kpt), dim = -1)
+
+                else:  # for YOLOv5 on AWS Inferentia https://github.com/ultralytics/yolov5/pull/2953
+                    xy = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                    wh = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i]  # wh
+                    if self.nkpt != 0:
+                        y[..., 6:] = (y[..., 6:] * 2. - 0.5 + self.grid[i].repeat((1,1,1,1,self.nkpt))) * self.stride[i]  # xy
+                    y = torch.cat((xy, wh, y[..., 4:]), -1)
+
+                z.append(y.view(bs, -1, self.no))
+
+        return x if self.training else (torch.cat(z, 1), x)
+
+    @staticmethod
+    def _make_grid(nx=20, ny=20):
+        yv, xv = torch.meshgrid([torch.arange(ny), torch.arange(nx)])
+        return torch.stack((xv, yv), 2).view((1, 1, ny, nx, 2)).float()
+
+
+class IAuxDetect(nn.Module):
+    stride = None  # strides computed during build
+    export = False  # onnx export
+    end2end = False
+    include_nms = False
+    concat = False
+
+    def __init__(self, nc=80, anchors=(), ch=()):  # detection layer
+        super(IAuxDetect, self).__init__()
+        self.nc = nc  # number of classes
+        self.no = nc + 5  # number of outputs per anchor
+        self.nl = len(anchors)  # number of detection layers
+        self.na = len(anchors[0]) // 2  # number of anchors
+        self.grid = [torch.zeros(1)] * self.nl  # init grid
+        a = torch.tensor(anchors).float().view(self.nl, -1, 2)
+        self.register_buffer('anchors', a)  # shape(nl,na,2)
+        self.register_buffer('anchor_grid', a.clone().view(self.nl, 1, -1, 1, 1, 2))  # shape(nl,1,na,1,1,2)
+        self.m = nn.ModuleList(nn.Conv2d(x, self.no * self.na, 1) for x in ch[:self.nl])  # output conv
+        self.m2 = nn.ModuleList(nn.Conv2d(x, self.no * self.na, 1) for x in ch[self.nl:])  # output conv
+        
+        self.ia = nn.ModuleList(ImplicitA(x) for x in ch[:self.nl])
+        self.im = nn.ModuleList(ImplicitM(self.no * self.na) for _ in ch[:self.nl])
+
+    def forward(self, x):
+        # x = x.copy()  # for profiling
+        z = []  # inference output
+        self.training |= self.export
+        for i in range(self.nl):
+            x[i] = self.m[i](self.ia[i](x[i]))  # conv
+            x[i] = self.im[i](x[i])
+            bs, _, ny, nx = x[i].shape  # x(bs,255,20,20) to x(bs,3,20,20,85)
+            x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+            
+            x[i+self.nl] = self.m2[i](x[i+self.nl])
+            x[i+self.nl] = x[i+self.nl].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+
+            if not self.training:  # inference
+                if self.grid[i].shape[2:4] != x[i].shape[2:4]:
+                    self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
+
+                y = x[i].sigmoid()
+                if not torch.onnx.is_in_onnx_export():
+                    y[..., 0:2] = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                    y[..., 2:4] = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i]  # wh
+                else:
+                    xy, wh, conf = y.split((2, 2, self.nc + 1), 4)  # y.tensor_split((2, 4, 5), 4)  # torch 1.8.0
+                    xy = xy * (2. * self.stride[i]) + (self.stride[i] * (self.grid[i] - 0.5))  # new xy
+                    wh = wh ** 2 * (4 * self.anchor_grid[i].data)  # new wh
+                    y = torch.cat((xy, wh, conf), 4)
+                z.append(y.view(bs, -1, self.no))
+
+        return x if self.training else (torch.cat(z, 1), x[:self.nl])
+
+    def fuseforward(self, x):
+        # x = x.copy()  # for profiling
+        z = []  # inference output
+        self.training |= self.export
+        for i in range(self.nl):
+            x[i] = self.m[i](x[i])  # conv
+            bs, _, ny, nx = x[i].shape  # x(bs,255,20,20) to x(bs,3,20,20,85)
+            x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+
+            if not self.training:  # inference
+                if self.grid[i].shape[2:4] != x[i].shape[2:4]:
+                    self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
+
+                y = x[i].sigmoid()
+                if not torch.onnx.is_in_onnx_export():
+                    y[..., 0:2] = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                    y[..., 2:4] = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i]  # wh
+                else:
+                    xy = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                    wh = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i].data  # wh
+                    y = torch.cat((xy, wh, y[..., 4:]), -1)
+                z.append(y.view(bs, -1, self.no))
+
+        if self.training:
+            out = x
+        elif self.end2end:
+            out = torch.cat(z, 1)
+        elif self.include_nms:
+            z = self.convert(z)
+            out = (z, )
+        elif self.concat:
+            out = torch.cat(z, 1)            
+        else:
+            out = (torch.cat(z, 1), x)
+
+        return out
+    
+    def fuse(self):
+        print("IAuxDetect.fuse")
+        # fuse ImplicitA and Convolution
+        for i in range(len(self.m)):
+            c1,c2,_,_ = self.m[i].weight.shape
+            c1_,c2_, _,_ = self.ia[i].implicit.shape
+            self.m[i].bias += torch.matmul(self.m[i].weight.reshape(c1,c2),self.ia[i].implicit.reshape(c2_,c1_)).squeeze(1)
+
+        # fuse ImplicitM and Convolution
+        for i in range(len(self.m)):
+            c1,c2, _,_ = self.im[i].implicit.shape
+            self.m[i].bias *= self.im[i].implicit.reshape(c2)
+            self.m[i].weight *= self.im[i].implicit.transpose(0,1)
+
+    @staticmethod
+    def _make_grid(nx=20, ny=20):
+        yv, xv = torch.meshgrid([torch.arange(ny), torch.arange(nx)])
+        return torch.stack((xv, yv), 2).view((1, 1, ny, nx, 2)).float()
+
+    def convert(self, z):
+        z = torch.cat(z, 1)
+        box = z[:, :, :4]
+        conf = z[:, :, 4:5]
+        score = z[:, :, 5:]
+        score *= conf
+        convert_matrix = torch.tensor([[1, 0, 1, 0], [0, 1, 0, 1], [-0.5, 0, 0.5, 0], [0, -0.5, 0, 0.5]],
+                                           dtype=torch.float32,
+                                           device=z.device)
+        box @= convert_matrix                          
+        return (box, score)
+
+
+class IBin(nn.Module):
+    stride = None  # strides computed during build
+    export = False  # onnx export
+
+    def __init__(self, nc=80, anchors=(), ch=(), bin_count=21):  # detection layer
+        super(IBin, self).__init__()
+        self.nc = nc  # number of classes
+        self.bin_count = bin_count
+
+        self.w_bin_sigmoid = SigmoidBin(bin_count=self.bin_count, min=0.0, max=4.0)
+        self.h_bin_sigmoid = SigmoidBin(bin_count=self.bin_count, min=0.0, max=4.0)
+        # classes, x,y,obj
+        self.no = nc + 3 + \
+            self.w_bin_sigmoid.get_length() + self.h_bin_sigmoid.get_length()   # w-bce, h-bce
+            # + self.x_bin_sigmoid.get_length() + self.y_bin_sigmoid.get_length()
+        
+        self.nl = len(anchors)  # number of detection layers
+        self.na = len(anchors[0]) // 2  # number of anchors
+        self.grid = [torch.zeros(1)] * self.nl  # init grid
+        a = torch.tensor(anchors).float().view(self.nl, -1, 2)
+        self.register_buffer('anchors', a)  # shape(nl,na,2)
+        self.register_buffer('anchor_grid', a.clone().view(self.nl, 1, -1, 1, 1, 2))  # shape(nl,1,na,1,1,2)
+        self.m = nn.ModuleList(nn.Conv2d(x, self.no * self.na, 1) for x in ch)  # output conv
+        
+        self.ia = nn.ModuleList(ImplicitA(x) for x in ch)
+        self.im = nn.ModuleList(ImplicitM(self.no * self.na) for _ in ch)
+
+    def forward(self, x):
+
+        #self.x_bin_sigmoid.use_fw_regression = True
+        #self.y_bin_sigmoid.use_fw_regression = True
+        self.w_bin_sigmoid.use_fw_regression = True
+        self.h_bin_sigmoid.use_fw_regression = True
+        
+        # x = x.copy()  # for profiling
+        z = []  # inference output
+        self.training |= self.export
+        for i in range(self.nl):
+            x[i] = self.m[i](self.ia[i](x[i]))  # conv
+            x[i] = self.im[i](x[i])
+            bs, _, ny, nx = x[i].shape  # x(bs,255,20,20) to x(bs,3,20,20,85)
+            x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
+
+            if not self.training:  # inference
+                if self.grid[i].shape[2:4] != x[i].shape[2:4]:
+                    self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
+
+                y = x[i].sigmoid()
+                y[..., 0:2] = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i]  # xy
+                #y[..., 2:4] = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i]  # wh
+                
+
+                #px = (self.x_bin_sigmoid.forward(y[..., 0:12]) + self.grid[i][..., 0]) * self.stride[i]
+                #py = (self.y_bin_sigmoid.forward(y[..., 12:24]) + self.grid[i][..., 1]) * self.stride[i]
+
+                pw = self.w_bin_sigmoid.forward(y[..., 2:24]) * self.anchor_grid[i][..., 0]
+                ph = self.h_bin_sigmoid.forward(y[..., 24:46]) * self.anchor_grid[i][..., 1]
+
+                #y[..., 0] = px
+                #y[..., 1] = py
+                y[..., 2] = pw
+                y[..., 3] = ph
+                
+                y = torch.cat((y[..., 0:4], y[..., 46:]), dim=-1)
+                
+                z.append(y.view(bs, -1, y.shape[-1]))
+
+        return x if self.training else (torch.cat(z, 1), x)
+
+    @staticmethod
+    def _make_grid(nx=20, ny=20):
+        yv, xv = torch.meshgrid([torch.arange(ny), torch.arange(nx)])
+        return torch.stack((xv, yv), 2).view((1, 1, ny, nx, 2)).float()
+
+
+class Model(nn.Module):
+    def __init__(self, cfg='yolor-csp-c.yaml', ch=3, nc=None, anchors=None):  # model, input channels, number of classes
+        super(Model, self).__init__()
+        self.traced = False
+        if isinstance(cfg, dict):
+            self.yaml = cfg  # model dict
+        else:  # is *.yaml
+            import yaml  # for torch hub
+            self.yaml_file = Path(cfg).name
+            with open(cfg) as f:
+                self.yaml = yaml.load(f, Loader=yaml.SafeLoader)  # model dict
+
+        # Define model
+        ch = self.yaml['ch'] = self.yaml.get('ch', ch)  # input channels
+        if nc and nc != self.yaml['nc']:
+            logger.info(f"Overriding model.yaml nc={self.yaml['nc']} with nc={nc}")
+            self.yaml['nc'] = nc  # override yaml value
+        if anchors:
+            logger.info(f'Overriding model.yaml anchors with anchors={anchors}')
+            self.yaml['anchors'] = round(anchors)  # override yaml value
+        self.model, self.save = parse_model(deepcopy(self.yaml), ch=[ch])  # model, savelist
+        self.names = [str(i) for i in range(self.yaml['nc'])]  # default names
+        # print([x.shape for x in self.forward(torch.zeros(1, ch, 64, 64))])
+
+        # Build strides, anchors
+        m = self.model[-1]  # Detect()
+        if isinstance(m, Detect):
+            s = 256  # 2x min stride
+            m.stride = torch.tensor([s / x.shape[-2] for x in self.forward(torch.zeros(1, ch, s, s))])  # forward
+            check_anchor_order(m)
+            m.anchors /= m.stride.view(-1, 1, 1)
+            self.stride = m.stride
+            self._initialize_biases()  # only run once
+            # print('Strides: %s' % m.stride.tolist())
+        if isinstance(m, IDetect):
+            s = 256  # 2x min stride
+            m.stride = torch.tensor([s / x.shape[-2] for x in self.forward(torch.zeros(1, ch, s, s))])  # forward
+            check_anchor_order(m)
+            m.anchors /= m.stride.view(-1, 1, 1)
+            self.stride = m.stride
+            self._initialize_biases()  # only run once
+            # print('Strides: %s' % m.stride.tolist())
+        if isinstance(m, IAuxDetect):
+            s = 256  # 2x min stride
+            m.stride = torch.tensor([s / x.shape[-2] for x in self.forward(torch.zeros(1, ch, s, s))[:4]])  # forward
+            #print(m.stride)
+            check_anchor_order(m)
+            m.anchors /= m.stride.view(-1, 1, 1)
+            self.stride = m.stride
+            self._initialize_aux_biases()  # only run once
+            # print('Strides: %s' % m.stride.tolist())
+        if isinstance(m, IBin):
+            s = 256  # 2x min stride
+            m.stride = torch.tensor([s / x.shape[-2] for x in self.forward(torch.zeros(1, ch, s, s))])  # forward
+            check_anchor_order(m)
+            m.anchors /= m.stride.view(-1, 1, 1)
+            self.stride = m.stride
+            self._initialize_biases_bin()  # only run once
+            # print('Strides: %s' % m.stride.tolist())
+        if isinstance(m, IKeypoint):
+            s = 256  # 2x min stride
+            m.stride = torch.tensor([s / x.shape[-2] for x in self.forward(torch.zeros(1, ch, s, s))])  # forward
+            check_anchor_order(m)
+            m.anchors /= m.stride.view(-1, 1, 1)
+            self.stride = m.stride
+            self._initialize_biases_kpt()  # only run once
+            # print('Strides: %s' % m.stride.tolist())
+
+        # Init weights, biases
+        initialize_weights(self)
+        self.info()
+        logger.info('')
+
+    def forward(self, x, augment=False, profile=False):
+        if augment:
+            img_size = x.shape[-2:]  # height, width
+            s = [1, 0.83, 0.67]  # scales
+            f = [None, 3, None]  # flips (2-ud, 3-lr)
+            y = []  # outputs
+            for si, fi in zip(s, f):
+                xi = scale_img(x.flip(fi) if fi else x, si, gs=int(self.stride.max()))
+                yi = self.forward_once(xi)[0]  # forward
+                # cv2.imwrite(f'img_{si}.jpg', 255 * xi[0].cpu().numpy().transpose((1, 2, 0))[:, :, ::-1])  # save
+                yi[..., :4] /= si  # de-scale
+                if fi == 2:
+                    yi[..., 1] = img_size[0] - yi[..., 1]  # de-flip ud
+                elif fi == 3:
+                    yi[..., 0] = img_size[1] - yi[..., 0]  # de-flip lr
+                y.append(yi)
+            return torch.cat(y, 1), None  # augmented inference, train
+        else:
+            return self.forward_once(x, profile)  # single-scale inference, train
+
+    def forward_once(self, x, profile=False):
+        y, dt = [], []  # outputs
+        for m in self.model:
+            if m.f != -1:  # if not from previous layer
+                x = y[m.f] if isinstance(m.f, int) else [x if j == -1 else y[j] for j in m.f]  # from earlier layers
+
+            if not hasattr(self, 'traced'):
+                self.traced=False
+
+            if self.traced:
+                if isinstance(m, Detect) or isinstance(m, IDetect) or isinstance(m, IAuxDetect) or isinstance(m, IKeypoint):
+                    break
+
+            if profile:
+                c = isinstance(m, (Detect, IDetect, IAuxDetect, IBin))
+                o = thop.profile(m, inputs=(x.copy() if c else x,), verbose=False)[0] / 1E9 * 2 if thop else 0  # FLOPS
+                for _ in range(10):
+                    m(x.copy() if c else x)
+                t = time_synchronized()
+                for _ in range(10):
+                    m(x.copy() if c else x)
+                dt.append((time_synchronized() - t) * 100)
+                print('%10.1f%10.0f%10.1fms %-40s' % (o, m.np, dt[-1], m.type))
+
+            x = m(x)  # run
+            
+            y.append(x if m.i in self.save else None)  # save output
+
+        if profile:
+            print('%.1fms total' % sum(dt))
+        return x
+
+    def _initialize_biases(self, cf=None):  # initialize biases into Detect(), cf is class frequency
+        # https://arxiv.org/abs/1708.02002 section 3.3
+        # cf = torch.bincount(torch.tensor(np.concatenate(dataset.labels, 0)[:, 0]).long(), minlength=nc) + 1.
+        m = self.model[-1]  # Detect() module
+        for mi, s in zip(m.m, m.stride):  # from
+            b = mi.bias.view(m.na, -1)  # conv.bias(255) to (3,85)
+            b.data[:, 4] += math.log(8 / (640 / s) ** 2)  # obj (8 objects per 640 image)
+            b.data[:, 5:] += math.log(0.6 / (m.nc - 0.99)) if cf is None else torch.log(cf / cf.sum())  # cls
+            mi.bias = torch.nn.Parameter(b.view(-1), requires_grad=True)
+
+    def _initialize_aux_biases(self, cf=None):  # initialize biases into Detect(), cf is class frequency
+        # https://arxiv.org/abs/1708.02002 section 3.3
+        # cf = torch.bincount(torch.tensor(np.concatenate(dataset.labels, 0)[:, 0]).long(), minlength=nc) + 1.
+        m = self.model[-1]  # Detect() module
+        for mi, mi2, s in zip(m.m, m.m2, m.stride):  # from
+            b = mi.bias.view(m.na, -1)  # conv.bias(255) to (3,85)
+            b.data[:, 4] += math.log(8 / (640 / s) ** 2)  # obj (8 objects per 640 image)
+            b.data[:, 5:] += math.log(0.6 / (m.nc - 0.99)) if cf is None else torch.log(cf / cf.sum())  # cls
+            mi.bias = torch.nn.Parameter(b.view(-1), requires_grad=True)
+            b2 = mi2.bias.view(m.na, -1)  # conv.bias(255) to (3,85)
+            b2.data[:, 4] += math.log(8 / (640 / s) ** 2)  # obj (8 objects per 640 image)
+            b2.data[:, 5:] += math.log(0.6 / (m.nc - 0.99)) if cf is None else torch.log(cf / cf.sum())  # cls
+            mi2.bias = torch.nn.Parameter(b2.view(-1), requires_grad=True)
+
+    def _initialize_biases_bin(self, cf=None):  # initialize biases into Detect(), cf is class frequency
+        # https://arxiv.org/abs/1708.02002 section 3.3
+        # cf = torch.bincount(torch.tensor(np.concatenate(dataset.labels, 0)[:, 0]).long(), minlength=nc) + 1.
+        m = self.model[-1]  # Bin() module
+        bc = m.bin_count
+        for mi, s in zip(m.m, m.stride):  # from
+            b = mi.bias.view(m.na, -1)  # conv.bias(255) to (3,85)
+            old = b[:, (0,1,2,bc+3)].data
+            obj_idx = 2*bc+4
+            b[:, :obj_idx].data += math.log(0.6 / (bc + 1 - 0.99))
+            b[:, obj_idx].data += math.log(8 / (640 / s) ** 2)  # obj (8 objects per 640 image)
+            b[:, (obj_idx+1):].data += math.log(0.6 / (m.nc - 0.99)) if cf is None else torch.log(cf / cf.sum())  # cls
+            b[:, (0,1,2,bc+3)].data = old
+            mi.bias = torch.nn.Parameter(b.view(-1), requires_grad=True)
+
+    def _initialize_biases_kpt(self, cf=None):  # initialize biases into Detect(), cf is class frequency
+        # https://arxiv.org/abs/1708.02002 section 3.3
+        # cf = torch.bincount(torch.tensor(np.concatenate(dataset.labels, 0)[:, 0]).long(), minlength=nc) + 1.
+        m = self.model[-1]  # Detect() module
+        for mi, s in zip(m.m, m.stride):  # from
+            b = mi.bias.view(m.na, -1)  # conv.bias(255) to (3,85)
+            b.data[:, 4] += math.log(8 / (640 / s) ** 2)  # obj (8 objects per 640 image)
+            b.data[:, 5:] += math.log(0.6 / (m.nc - 0.99)) if cf is None else torch.log(cf / cf.sum())  # cls
+            mi.bias = torch.nn.Parameter(b.view(-1), requires_grad=True)
+
+    def _print_biases(self):
+        m = self.model[-1]  # Detect() module
+        for mi in m.m:  # from
+            b = mi.bias.detach().view(m.na, -1).T  # conv.bias(255) to (3,85)
+            print(('%6g Conv2d.bias:' + '%10.3g' * 6) % (mi.weight.shape[1], *b[:5].mean(1).tolist(), b[5:].mean()))
+
+    # def _print_weights(self):
+    #     for m in self.model.modules():
+    #         if type(m) is Bottleneck:
+    #             print('%10.3g' % (m.w.detach().sigmoid() * 2))  # shortcut weights
+
+    def fuse(self):  # fuse model Conv2d() + BatchNorm2d() layers
+        print('Fusing layers... ')
+        for m in self.model.modules():
+            if isinstance(m, RepConv):
+                #print(f" fuse_repvgg_block")
+                m.fuse_repvgg_block()
+            elif isinstance(m, RepConv_OREPA):
+                #print(f" switch_to_deploy")
+                m.switch_to_deploy()
+            elif type(m) is Conv and hasattr(m, 'bn'):
+                m.conv = fuse_conv_and_bn(m.conv, m.bn)  # update conv
+                delattr(m, 'bn')  # remove batchnorm
+                m.forward = m.fuseforward  # update forward
+            elif isinstance(m, (IDetect, IAuxDetect)):
+                m.fuse()
+                m.forward = m.fuseforward
+        self.info()
+        return self
+
+    def nms(self, mode=True):  # add or remove NMS module
+        present = type(self.model[-1]) is NMS  # last layer is NMS
+        if mode and not present:
+            print('Adding NMS... ')
+            m = NMS()  # module
+            m.f = -1  # from
+            m.i = self.model[-1].i + 1  # index
+            self.model.add_module(name='%s' % m.i, module=m)  # add
+            self.eval()
+        elif not mode and present:
+            print('Removing NMS... ')
+            self.model = self.model[:-1]  # remove
+        return self
+
+    def autoshape(self):  # add autoShape module
+        print('Adding autoShape... ')
+        m = autoShape(self)  # wrap model
+        copy_attr(m, self, include=('yaml', 'nc', 'hyp', 'names', 'stride'), exclude=())  # copy attributes
+        return m
+
+    def info(self, verbose=False, img_size=640):  # print model information
+        model_info(self, verbose, img_size)
+
+
+def parse_model(d, ch):  # model_dict, input_channels(3)
+    logger.info('\n%3s%18s%3s%10s  %-40s%-30s' % ('', 'from', 'n', 'params', 'module', 'arguments'))
+    anchors, nc, gd, gw = d['anchors'], d['nc'], d['depth_multiple'], d['width_multiple']
+    na = (len(anchors[0]) // 2) if isinstance(anchors, list) else anchors  # number of anchors
+    no = na * (nc + 5)  # number of outputs = anchors * (classes + 5)
+
+    layers, save, c2 = [], [], ch[-1]  # layers, savelist, ch out
+    for i, (f, n, m, args) in enumerate(d['backbone'] + d['head']):  # from, number, module, args
+        m = eval(m) if isinstance(m, str) else m  # eval strings
+        for j, a in enumerate(args):
+            try:
+                args[j] = eval(a) if isinstance(a, str) else a  # eval strings
+            except:
+                pass
+
+        n = max(round(n * gd), 1) if n > 1 else n  # depth gain
+        if m in [nn.Conv2d, Conv, RobustConv, RobustConv2, DWConv, GhostConv, RepConv, RepConv_OREPA, DownC, 
+                 SPP, SPPF, SPPCSPC, GhostSPPCSPC, MixConv2d, Focus, Stem, GhostStem, CrossConv, 
+                 Bottleneck, BottleneckCSPA, BottleneckCSPB, BottleneckCSPC, 
+                 RepBottleneck, RepBottleneckCSPA, RepBottleneckCSPB, RepBottleneckCSPC,  
+                 Res, ResCSPA, ResCSPB, ResCSPC, 
+                 RepRes, RepResCSPA, RepResCSPB, RepResCSPC, 
+                 ResX, ResXCSPA, ResXCSPB, ResXCSPC, 
+                 RepResX, RepResXCSPA, RepResXCSPB, RepResXCSPC, 
+                 Ghost, GhostCSPA, GhostCSPB, GhostCSPC,
+                 SwinTransformerBlock, STCSPA, STCSPB, STCSPC,
+                 SwinTransformer2Block, ST2CSPA, ST2CSPB, ST2CSPC]:
+            c1, c2 = ch[f], args[0]
+            if c2 != no:  # if not output
+                c2 = make_divisible(c2 * gw, 8)
+
+            args = [c1, c2, *args[1:]]
+            if m in [DownC, SPPCSPC, GhostSPPCSPC, 
+                     BottleneckCSPA, BottleneckCSPB, BottleneckCSPC, 
+                     RepBottleneckCSPA, RepBottleneckCSPB, RepBottleneckCSPC, 
+                     ResCSPA, ResCSPB, ResCSPC, 
+                     RepResCSPA, RepResCSPB, RepResCSPC, 
+                     ResXCSPA, ResXCSPB, ResXCSPC, 
+                     RepResXCSPA, RepResXCSPB, RepResXCSPC,
+                     GhostCSPA, GhostCSPB, GhostCSPC,
+                     STCSPA, STCSPB, STCSPC,
+                     ST2CSPA, ST2CSPB, ST2CSPC]:
+                args.insert(2, n)  # number of repeats
+                n = 1
+        elif m is nn.BatchNorm2d:
+            args = [ch[f]]
+        elif m is Concat:
+            c2 = sum([ch[x] for x in f])
+        elif m is Chuncat:
+            c2 = sum([ch[x] for x in f])
+        elif m is Shortcut:
+            c2 = ch[f[0]]
+        elif m is Foldcut:
+            c2 = ch[f] // 2
+        elif m in [Detect, IDetect, IAuxDetect, IBin, IKeypoint]:
+            args.append([ch[x] for x in f])
+            if isinstance(args[1], int):  # number of anchors
+                args[1] = [list(range(args[1] * 2))] * len(f)
+        elif m is ReOrg:
+            c2 = ch[f] * 4
+        elif m is Contract:
+            c2 = ch[f] * args[0] ** 2
+        elif m is Expand:
+            c2 = ch[f] // args[0] ** 2
+        else:
+            c2 = ch[f]
+
+        m_ = nn.Sequential(*[m(*args) for _ in range(n)]) if n > 1 else m(*args)  # module
+        t = str(m)[8:-2].replace('__main__.', '')  # module type
+        np = sum([x.numel() for x in m_.parameters()])  # number params
+        m_.i, m_.f, m_.type, m_.np = i, f, t, np  # attach index, 'from' index, type, number params
+        logger.info('%3s%18s%3s%10.0f  %-40s%-30s' % (i, f, n, np, t, args))  # print
+        save.extend(x % i for x in ([f] if isinstance(f, int) else f) if x != -1)  # append to savelist
+        layers.append(m_)
+        if i == 0:
+            ch = []
+        ch.append(c2)
+    return nn.Sequential(*layers), sorted(save)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--cfg', type=str, default='yolor-csp-c.yaml', help='model.yaml')
+    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
+    parser.add_argument('--profile', action='store_true', help='profile model speed')
+    opt = parser.parse_args()
+    opt.cfg = check_file(opt.cfg)  # check file
+    set_logging()
+    device = select_device(opt.device)
+
+    # Create model
+    model = Model(opt.cfg).to(device)
+    model.train()
+    
+    if opt.profile:
+        img = torch.rand(1, 3, 640, 640).to(device)
+        y = model(img, profile=True)
+
+    # Profile
+    # img = torch.rand(8 if torch.cuda.is_available() else 1, 3, 640, 640).to(device)
+    # y = model(img, profile=True)
+
+    # Tensorboard
+    # from torch.utils.tensorboard import SummaryWriter
+    # tb_writer = SummaryWriter()
+    # print("Run 'tensorboard --logdir=models/runs' to view tensorboard at http://localhost:6006/")
+    # tb_writer.add_graph(model.model, img)  # add model to tensorboard
+    # tb_writer.add_image('test', img[0], dataformats='CWH')  # add model to tensorboard