Duplicate from InpaintAI/Inpaint-Anything

Co-authored-by: RunsenFeng <RysonFeng@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,39 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+**.png filter=lfs diff=lfs merge=lfs -text
+**.gif filter=lfs diff=lfs merge=lfs -text
+**.jpg filter=lfs diff=lfs merge=lfs -text
+third_party/** filter=lfs diff=lfs merge=lfs -text
+pretrained_models/** filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,12 @@
+# Python byte code, etc.
+__pycache__/
+
+# C/C++ object files/libraries
+*.o
+*.so
+
+# macOS
+**/.DS_Store
+
+# tmp
+~*

LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -0,0 +1,14 @@
+---
+title: Inpaint Anything
+emoji: ⚡
+colorFrom: red
+colorTo: red
+sdk: gradio
+sdk_version: 3.27.0
+app_file: app.py
+pinned: false
+license: apache-2.0
+duplicated_from: InpaintAI/Inpaint-Anything
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,231 @@
+import os
+import sys
+# sys.path.append(os.path.abspath(os.path.dirname(os.getcwd())))
+# os.chdir("../")
+import gradio as gr
+import numpy as np
+from pathlib import Path
+from matplotlib import pyplot as plt
+import torch
+import tempfile
+from lama_inpaint import inpaint_img_with_lama, build_lama_model, inpaint_img_with_builded_lama
+from utils import load_img_to_array, save_array_to_img, dilate_mask, \
+    show_mask, show_points
+from PIL import Image
+sys.path.insert(0, str(Path(__file__).resolve().parent / "third_party" / "segment-anything"))
+from segment_anything import SamPredictor, sam_model_registry
+import argparse
+
+def setup_args(parser):
+    parser.add_argument(
+        "--lama_config", type=str,
+        default="./third_party/lama/configs/prediction/default.yaml",
+        help="The path to the config file of lama model. "
+             "Default: the config of big-lama",
+    )
+    parser.add_argument(
+        "--lama_ckpt", type=str,
+        default="pretrained_models/big-lama",
+        help="The path to the lama checkpoint.",
+    )
+    parser.add_argument(
+        "--sam_ckpt", type=str,
+        default="./pretrained_models/sam_vit_h_4b8939.pth",
+        help="The path to the SAM checkpoint to use for mask generation.",
+    )
+def mkstemp(suffix, dir=None):
+    fd, path = tempfile.mkstemp(suffix=f"{suffix}", dir=dir)
+    os.close(fd)
+    return Path(path)
+
+
+def get_sam_feat(img):
+    model['sam'].set_image(img)
+    features = model['sam'].features 
+    orig_h = model['sam'].orig_h 
+    orig_w = model['sam'].orig_w 
+    input_h = model['sam'].input_h 
+    input_w = model['sam'].input_w 
+    model['sam'].reset_image()
+    return features, orig_h, orig_w, input_h, input_w
+
+ 
+def get_masked_img(img, w, h, features, orig_h, orig_w, input_h, input_w, dilate_kernel_size):
+    point_coords = [w, h]
+    point_labels = [1]
+
+    model['sam'].is_image_set = True
+    model['sam'].features = features
+    model['sam'].orig_h = orig_h
+    model['sam'].orig_w = orig_w
+    model['sam'].input_h = input_h
+    model['sam'].input_w = input_w
+
+    # model['sam'].set_image(img) # todo : update here for accelerating
+    masks, _, _ = model['sam'].predict(
+        point_coords=np.array([point_coords]),
+        point_labels=np.array(point_labels),
+        multimask_output=True,
+    )
+
+    masks = masks.astype(np.uint8) * 255
+
+    # dilate mask to avoid unmasked edge effect
+    if dilate_kernel_size is not None:
+        masks = [dilate_mask(mask, dilate_kernel_size) for mask in masks]
+    else:
+        masks = [mask for mask in masks]
+
+    figs = []
+    for idx, mask in enumerate(masks):
+        # save the pointed and masked image
+        tmp_p = mkstemp(".png")
+        dpi = plt.rcParams['figure.dpi']
+        height, width = img.shape[:2]
+        fig = plt.figure(figsize=(width/dpi/0.77, height/dpi/0.77))
+        plt.imshow(img)
+        plt.axis('off')
+        show_points(plt.gca(), [point_coords], point_labels,
+                    size=(width*0.04)**2)
+        show_mask(plt.gca(), mask, random_color=False)
+        plt.tight_layout()
+        plt.savefig(tmp_p, bbox_inches='tight', pad_inches=0)
+        figs.append(fig)
+        plt.close()
+    return *figs, *masks
+
+
+def get_inpainted_img(img, mask0, mask1, mask2):
+    lama_config = args.lama_config
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    out = []
+    for mask in [mask0, mask1, mask2]:
+        if len(mask.shape)==3:
+            mask = mask[:,:,0]
+        img_inpainted = inpaint_img_with_builded_lama(
+            model['lama'], img, mask, lama_config, device=device)
+        out.append(img_inpainted)
+    return out
+
+
+# get args 
+parser = argparse.ArgumentParser()
+setup_args(parser)
+args = parser.parse_args(sys.argv[1:])
+# build models
+model = {}
+# build the sam model
+model_type="vit_h"
+ckpt_p=args.sam_ckpt
+model_sam = sam_model_registry[model_type](checkpoint=ckpt_p)
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model_sam.to(device=device)
+model['sam'] = SamPredictor(model_sam)
+
+# build the lama model
+lama_config = args.lama_config
+lama_ckpt = args.lama_ckpt
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model['lama'] = build_lama_model(lama_config, lama_ckpt, device=device)
+
+button_size = (100,50)
+with gr.Blocks() as demo:
+    features = gr.State(None)
+    orig_h = gr.State(None)
+    orig_w = gr.State(None)
+    input_h = gr.State(None)
+    input_w = gr.State(None)
+
+    with gr.Row().style(mobile_collapse=False, equal_height=True):
+        with gr.Column(variant="panel"):
+            with gr.Row():
+                gr.Markdown("## Input Image")
+            with gr.Row():
+                img = gr.Image(label="Input Image").style(height="200px")
+        with gr.Column(variant="panel"):
+            with gr.Row():
+                gr.Markdown("## Pointed Image")
+            with gr.Row():
+                img_pointed = gr.Plot(label='Pointed Image')
+        with gr.Column(variant="panel"):
+            with gr.Row():
+                gr.Markdown("## Control Panel")
+            with gr.Row():
+                w = gr.Number(label="Point Coordinate W")
+                h = gr.Number(label="Point Coordinate H")
+            dilate_kernel_size = gr.Slider(label="Dilate Kernel Size", minimum=0, maximum=100, step=1, value=15)
+            sam_mask = gr.Button("Predict Mask", variant="primary").style(full_width=True, size="sm")
+            lama = gr.Button("Inpaint Image", variant="primary").style(full_width=True, size="sm")
+            clear_button_image = gr.Button(value="Reset", label="Reset", variant="secondary").style(full_width=True, size="sm")
+
+    # todo: maybe we can delete this row, for it's unnecessary to show the original mask for customers
+    with gr.Row(variant="panel"):
+        with gr.Column():
+            with gr.Row():
+                gr.Markdown("## Segmentation Mask")
+            with gr.Row():
+                mask_0 = gr.outputs.Image(type="numpy", label="Segmentation Mask 0").style(height="200px")
+                mask_1 = gr.outputs.Image(type="numpy", label="Segmentation Mask 1").style(height="200px")
+                mask_2 = gr.outputs.Image(type="numpy", label="Segmentation Mask 2").style(height="200px")
+
+    with gr.Row(variant="panel"):
+        with gr.Column():
+            with gr.Row():
+                gr.Markdown("## Image with Mask")
+            with gr.Row():
+                img_with_mask_0 = gr.Plot(label="Image with Segmentation Mask 0")
+                img_with_mask_1 = gr.Plot(label="Image with Segmentation Mask 1")
+                img_with_mask_2 = gr.Plot(label="Image with Segmentation Mask 2")
+
+    with gr.Row(variant="panel"):
+        with gr.Column():
+            with gr.Row():
+                gr.Markdown("## Image Removed with Mask")
+            with gr.Row():
+                img_rm_with_mask_0 = gr.outputs.Image(
+                    type="numpy", label="Image Removed with Segmentation Mask 0").style(height="200px")
+                img_rm_with_mask_1 = gr.outputs.Image(
+                    type="numpy", label="Image Removed with Segmentation Mask 1").style(height="200px")
+                img_rm_with_mask_2 = gr.outputs.Image(
+                    type="numpy", label="Image Removed with Segmentation Mask 2").style(height="200px")
+
+
+    def get_select_coords(img, evt: gr.SelectData):
+        dpi = plt.rcParams['figure.dpi']
+        height, width = img.shape[:2]
+        fig = plt.figure(figsize=(width/dpi/0.77, height/dpi/0.77))
+        plt.imshow(img)
+        plt.axis('off')
+        plt.tight_layout()
+        show_points(plt.gca(), [[evt.index[0], evt.index[1]]], [1],
+                    size=(width*0.04)**2)
+        return evt.index[0], evt.index[1], fig
+
+    img.select(get_select_coords, [img], [w, h, img_pointed])
+    img.upload(get_sam_feat, [img], [features, orig_h, orig_w, input_h, input_w])
+
+    sam_mask.click(
+        get_masked_img,
+        [img, w, h, features, orig_h, orig_w, input_h, input_w, dilate_kernel_size],
+        [img_with_mask_0, img_with_mask_1, img_with_mask_2, mask_0, mask_1, mask_2]
+    )
+
+    lama.click(
+        get_inpainted_img,
+        [img, mask_0, mask_1, mask_2],
+        [img_rm_with_mask_0, img_rm_with_mask_1, img_rm_with_mask_2]
+    )
+
+
+    def reset(*args):
+        return [None for _ in args]
+
+    clear_button_image.click(
+        reset,
+        [img, features, img_pointed, w, h, mask_0, mask_1, mask_2, img_with_mask_0, img_with_mask_1, img_with_mask_2, img_rm_with_mask_0, img_rm_with_mask_1, img_rm_with_mask_2],
+        [img, features, img_pointed, w, h, mask_0, mask_1, mask_2, img_with_mask_0, img_with_mask_1, img_with_mask_2, img_rm_with_mask_0, img_rm_with_mask_1, img_rm_with_mask_2]
+    )
+
+if __name__ == "__main__":
+    demo.launch()
+    

fill_anything.py

@@ -0,0 +1,128 @@
+import cv2
+import sys
+import argparse
+import numpy as np
+import torch
+from pathlib import Path
+from matplotlib import pyplot as plt
+from typing import Any, Dict, List
+
+from sam_segment import predict_masks_with_sam
+from stable_diffusion_inpaint import fill_img_with_sd
+from utils import load_img_to_array, save_array_to_img, dilate_mask, \
+    show_mask, show_points
+
+
+def setup_args(parser):
+    parser.add_argument(
+        "--input_img", type=str, required=True,
+        help="Path to a single input img",
+    )
+    parser.add_argument(
+        "--point_coords", type=float, nargs='+', required=True,
+        help="The coordinate of the point prompt, [coord_W coord_H].",
+    )
+    parser.add_argument(
+        "--point_labels", type=int, nargs='+', required=True,
+        help="The labels of the point prompt, 1 or 0.",
+    )
+    parser.add_argument(
+        "--text_prompt", type=str, required=True,
+        help="Text prompt",
+    )
+    parser.add_argument(
+        "--dilate_kernel_size", type=int, default=None,
+        help="Dilate kernel size. Default: None",
+    )
+    parser.add_argument(
+        "--output_dir", type=str, required=True,
+        help="Output path to the directory with results.",
+    )
+    parser.add_argument(
+        "--sam_model_type", type=str,
+        default="vit_h", choices=['vit_h', 'vit_l', 'vit_b'],
+        help="The type of sam model to load. Default: 'vit_h"
+    )
+    parser.add_argument(
+        "--sam_ckpt", type=str, required=True,
+        help="The path to the SAM checkpoint to use for mask generation.",
+    )
+    parser.add_argument(
+        "--seed", type=int,
+        help="Specify seed for reproducibility.",
+    )
+    parser.add_argument(
+        "--deterministic", action="store_true",
+        help="Use deterministic algorithms for reproducibility.",
+    )
+
+
+
+if __name__ == "__main__":
+    """Example usage:
+    python fill_anything.py \
+        --input_img FA_demo/FA1_dog.png \
+        --point_coords 750 500 \
+        --point_labels 1 \
+        --text_prompt "a teddy bear on a bench" \
+        --dilate_kernel_size 15 \
+        --output_dir ./results \
+        --sam_model_type "vit_h" \
+        --sam_ckpt sam_vit_h_4b8939.pth 
+    """
+    parser = argparse.ArgumentParser()
+    setup_args(parser)
+    args = parser.parse_args(sys.argv[1:])
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    img = load_img_to_array(args.input_img)
+
+    masks, _, _ = predict_masks_with_sam(
+        img,
+        [args.point_coords],
+        args.point_labels,
+        model_type=args.sam_model_type,
+        ckpt_p=args.sam_ckpt,
+        device=device,
+    )
+    masks = masks.astype(np.uint8) * 255
+
+    # dilate mask to avoid unmasked edge effect
+    if args.dilate_kernel_size is not None:
+        masks = [dilate_mask(mask, args.dilate_kernel_size) for mask in masks]
+
+    # visualize the segmentation results
+    img_stem = Path(args.input_img).stem
+    out_dir = Path(args.output_dir) / img_stem
+    out_dir.mkdir(parents=True, exist_ok=True)
+    for idx, mask in enumerate(masks):
+        # path to the results
+        mask_p = out_dir / f"mask_{idx}.png"
+        img_points_p = out_dir / f"with_points.png"
+        img_mask_p = out_dir / f"with_{Path(mask_p).name}"
+
+        # save the mask
+        save_array_to_img(mask, mask_p)
+
+        # save the pointed and masked image
+        dpi = plt.rcParams['figure.dpi']
+        height, width = img.shape[:2]
+        plt.figure(figsize=(width/dpi/0.77, height/dpi/0.77))
+        plt.imshow(img)
+        plt.axis('off')
+        show_points(plt.gca(), [args.point_coords], args.point_labels,
+                    size=(width*0.04)**2)
+        plt.savefig(img_points_p, bbox_inches='tight', pad_inches=0)
+        show_mask(plt.gca(), mask, random_color=False)
+        plt.savefig(img_mask_p, bbox_inches='tight', pad_inches=0)
+        plt.close()
+
+    # fill the masked image
+    for idx, mask in enumerate(masks):
+        if args.seed is not None:
+            torch.manual_seed(args.seed)
+        mask_p = out_dir / f"mask_{idx}.png"
+        img_filled_p = out_dir / f"filled_with_{Path(mask_p).name}"
+        img_filled = fill_img_with_sd(
+            img, mask, args.text_prompt, device=device)
+        save_array_to_img(img_filled, img_filled_p)

lama_inpaint.py

@@ -0,0 +1,205 @@
+import os
+import sys
+import numpy as np
+import torch
+import yaml
+import glob
+import argparse
+from PIL import Image
+from omegaconf import OmegaConf
+from pathlib import Path
+
+os.environ['OMP_NUM_THREADS'] = '1'
+os.environ['OPENBLAS_NUM_THREADS'] = '1'
+os.environ['MKL_NUM_THREADS'] = '1'
+os.environ['VECLIB_MAXIMUM_THREADS'] = '1'
+os.environ['NUMEXPR_NUM_THREADS'] = '1'
+
+sys.path.insert(0, str(Path(__file__).resolve().parent / "third_party" / "lama"))
+
+from saicinpainting.evaluation.utils import move_to_device
+from saicinpainting.training.trainers import load_checkpoint
+from saicinpainting.evaluation.data import pad_tensor_to_modulo
+
+from utils import load_img_to_array, save_array_to_img
+
+
+@torch.no_grad()
+def inpaint_img_with_lama(
+        img: np.ndarray,
+        mask: np.ndarray,
+        config_p: str,
+        ckpt_p: str,
+        mod=8,
+        device="cuda"
+):
+    assert len(mask.shape) == 2
+    if np.max(mask) == 1:
+        mask = mask * 255
+    img = torch.from_numpy(img).float().div(255.)
+    mask = torch.from_numpy(mask).float()
+    predict_config = OmegaConf.load(config_p)
+    predict_config.model.path = ckpt_p
+    # device = torch.device(predict_config.device)
+    device = torch.device(device)
+
+    train_config_path = os.path.join(
+        predict_config.model.path, 'config.yaml')
+
+    with open(train_config_path, 'r') as f:
+        train_config = OmegaConf.create(yaml.safe_load(f))
+
+    train_config.training_model.predict_only = True
+    train_config.visualizer.kind = 'noop'
+
+    checkpoint_path = os.path.join(
+        predict_config.model.path, 'models',
+        predict_config.model.checkpoint
+    )
+    model = load_checkpoint(
+        train_config, checkpoint_path, strict=False, map_location=device)
+    model.freeze()
+    if not predict_config.get('refine', False):
+        model.to(device)
+
+    batch = {}
+    batch['image'] = img.permute(2, 0, 1).unsqueeze(0)
+    batch['mask'] = mask[None, None]
+    unpad_to_size = [batch['image'].shape[2], batch['image'].shape[3]]
+    batch['image'] = pad_tensor_to_modulo(batch['image'], mod)
+    batch['mask'] = pad_tensor_to_modulo(batch['mask'], mod)
+    batch = move_to_device(batch, device)
+    batch['mask'] = (batch['mask'] > 0) * 1
+
+    batch = model(batch)
+    cur_res = batch[predict_config.out_key][0].permute(1, 2, 0)
+    cur_res = cur_res.detach().cpu().numpy()
+
+    if unpad_to_size is not None:
+        orig_height, orig_width = unpad_to_size
+        cur_res = cur_res[:orig_height, :orig_width]
+
+    cur_res = np.clip(cur_res * 255, 0, 255).astype('uint8')
+    return cur_res
+
+
+def build_lama_model(        
+        config_p: str,
+        ckpt_p: str,
+        device="cuda"
+):
+    predict_config = OmegaConf.load(config_p)
+    predict_config.model.path = ckpt_p
+    # device = torch.device(predict_config.device)
+    device = torch.device(device)
+
+    train_config_path = os.path.join(
+        predict_config.model.path, 'config.yaml')
+
+    with open(train_config_path, 'r') as f:
+        train_config = OmegaConf.create(yaml.safe_load(f))
+
+    train_config.training_model.predict_only = True
+    train_config.visualizer.kind = 'noop'
+
+    checkpoint_path = os.path.join(
+        predict_config.model.path, 'models',
+        predict_config.model.checkpoint
+    )
+    model = load_checkpoint(
+        train_config, checkpoint_path, strict=False, map_location=device)
+    model.freeze()
+    if not predict_config.get('refine', False):
+        model.to(device)
+
+    return model
+
+
+@torch.no_grad()
+def inpaint_img_with_builded_lama(
+        model,
+        img: np.ndarray,
+        mask: np.ndarray,
+        config_p: str,
+        mod=8,
+        device="cuda"
+):
+    assert len(mask.shape) == 2
+    if np.max(mask) == 1:
+        mask = mask * 255
+    img = torch.from_numpy(img).float().div(255.)
+    mask = torch.from_numpy(mask).float()
+    predict_config = OmegaConf.load(config_p)
+
+    batch = {}
+    batch['image'] = img.permute(2, 0, 1).unsqueeze(0)
+    batch['mask'] = mask[None, None]
+    unpad_to_size = [batch['image'].shape[2], batch['image'].shape[3]]
+    batch['image'] = pad_tensor_to_modulo(batch['image'], mod)
+    batch['mask'] = pad_tensor_to_modulo(batch['mask'], mod)
+    batch = move_to_device(batch, device)
+    batch['mask'] = (batch['mask'] > 0) * 1
+
+    batch = model(batch)
+    cur_res = batch[predict_config.out_key][0].permute(1, 2, 0)
+    cur_res = cur_res.detach().cpu().numpy()
+
+    if unpad_to_size is not None:
+        orig_height, orig_width = unpad_to_size
+        cur_res = cur_res[:orig_height, :orig_width]
+
+    cur_res = np.clip(cur_res * 255, 0, 255).astype('uint8')
+    return cur_res
+
+
+def setup_args(parser):
+    parser.add_argument(
+        "--input_img", type=str, required=True,
+        help="Path to a single input img",
+    )
+    parser.add_argument(
+        "--input_mask_glob", type=str, required=True,
+        help="Glob to input masks",
+    )
+    parser.add_argument(
+        "--output_dir", type=str, required=True,
+        help="Output path to the directory with results.",
+    )
+    parser.add_argument(
+        "--lama_config", type=str,
+        default="./third_party/lama/configs/prediction/default.yaml",
+        help="The path to the config file of lama model. "
+             "Default: the config of big-lama",
+    )
+    parser.add_argument(
+        "--lama_ckpt", type=str, required=True,
+        help="The path to the lama checkpoint.",
+    )
+
+
+if __name__ == "__main__":
+    """Example usage:
+    python lama_inpaint.py \
+        --input_img FA_demo/FA1_dog.png \
+        --input_mask_glob "results/FA1_dog/mask*.png" \
+        --output_dir results \
+        --lama_config lama/configs/prediction/default.yaml \
+        --lama_ckpt big-lama 
+    """
+    parser = argparse.ArgumentParser()
+    setup_args(parser)
+    args = parser.parse_args(sys.argv[1:])
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    img_stem = Path(args.input_img).stem
+    mask_ps = sorted(glob.glob(args.input_mask_glob))
+    out_dir = Path(args.output_dir) / img_stem
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    img = load_img_to_array(args.input_img)
+    for mask_p in mask_ps:
+        mask = load_img_to_array(mask_p)
+        img_inpainted_p = out_dir / f"inpainted_with_{Path(mask_p).name}"
+        img_inpainted = inpaint_img_with_lama(
+            img, mask, args.lama_config, args.lama_ckpt, device=device)
+        save_array_to_img(img_inpainted, img_inpainted_p)

pretrained_models/.gitkeep

@@ -0,0 +1 @@
+# Put pretrained models into this directory

pretrained_models/big-lama/config.yaml

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fdeed49926e13b101c4dd9e193acec9e58677dfdb4ba49dd6a3a8927964e2a7
+size 3947

pretrained_models/big-lama/models/best.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fccb7adffd53ec0974ee5503c3731c2c2f1e7e07856fd9228cdcc0b46fd5d423
+size 410046389

pretrained_models/sam_vit_h_4b8939.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7bf3b02f3ebf1267aba913ff637d9a2d5c33d3173bb679e46d9f338c26f262e
+size 2564550879

remove_anything.py

@@ -0,0 +1,122 @@
+import torch
+import sys
+import argparse
+import numpy as np
+from pathlib import Path
+from matplotlib import pyplot as plt
+
+from sam_segment import predict_masks_with_sam
+from lama_inpaint import inpaint_img_with_lama
+from utils import load_img_to_array, save_array_to_img, dilate_mask, \
+    show_mask, show_points
+
+
+def setup_args(parser):
+    parser.add_argument(
+        "--input_img", type=str, required=True,
+        help="Path to a single input img",
+    )
+    parser.add_argument(
+        "--point_coords", type=float, nargs='+', required=True,
+        help="The coordinate of the point prompt, [coord_W coord_H].",
+    )
+    parser.add_argument(
+        "--point_labels", type=int, nargs='+', required=True,
+        help="The labels of the point prompt, 1 or 0.",
+    )
+    parser.add_argument(
+        "--dilate_kernel_size", type=int, default=None,
+        help="Dilate kernel size. Default: None",
+    )
+    parser.add_argument(
+        "--output_dir", type=str, required=True,
+        help="Output path to the directory with results.",
+    )
+    parser.add_argument(
+        "--sam_model_type", type=str,
+        default="vit_h", choices=['vit_h', 'vit_l', 'vit_b'],
+        help="The type of sam model to load. Default: 'vit_h"
+    )
+    parser.add_argument(
+        "--sam_ckpt", type=str, required=True,
+        help="The path to the SAM checkpoint to use for mask generation.",
+    )
+    parser.add_argument(
+        "--lama_config", type=str,
+        default="./lama/configs/prediction/default.yaml",
+        help="The path to the config file of lama model. "
+             "Default: the config of big-lama",
+    )
+    parser.add_argument(
+        "--lama_ckpt", type=str, required=True,
+        help="The path to the lama checkpoint.",
+    )
+
+
+if __name__ == "__main__":
+    """Example usage:
+    python remove_anything.py \
+        --input_img FA_demo/FA1_dog.png \
+        --point_coords 750 500 \
+        --point_labels 1 \
+        --dilate_kernel_size 15 \
+        --output_dir ./results \
+        --sam_model_type "vit_h" \
+        --sam_ckpt sam_vit_h_4b8939.pth \
+        --lama_config lama/configs/prediction/default.yaml \
+        --lama_ckpt big-lama 
+    """
+    parser = argparse.ArgumentParser()
+    setup_args(parser)
+    args = parser.parse_args(sys.argv[1:])
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    img = load_img_to_array(args.input_img)
+
+    masks, _, _ = predict_masks_with_sam(
+        img,
+        [args.point_coords],
+        args.point_labels,
+        model_type=args.sam_model_type,
+        ckpt_p=args.sam_ckpt,
+        device=device,
+    )
+    masks = masks.astype(np.uint8) * 255
+
+    # dilate mask to avoid unmasked edge effect
+    if args.dilate_kernel_size is not None:
+        masks = [dilate_mask(mask, args.dilate_kernel_size) for mask in masks]
+
+    # visualize the segmentation results
+    img_stem = Path(args.input_img).stem
+    out_dir = Path(args.output_dir) / img_stem
+    out_dir.mkdir(parents=True, exist_ok=True)
+    for idx, mask in enumerate(masks):
+        # path to the results
+        mask_p = out_dir / f"mask_{idx}.png"
+        img_points_p = out_dir / f"with_points.png"
+        img_mask_p = out_dir / f"with_{Path(mask_p).name}"
+
+        # save the mask
+        save_array_to_img(mask, mask_p)
+
+        # save the pointed and masked image
+        dpi = plt.rcParams['figure.dpi']
+        height, width = img.shape[:2]
+        plt.figure(figsize=(width/dpi/0.77, height/dpi/0.77))
+        plt.imshow(img)
+        plt.axis('off')
+        show_points(plt.gca(), [args.point_coords], args.point_labels,
+                    size=(width*0.04)**2)
+        plt.savefig(img_points_p, bbox_inches='tight', pad_inches=0)
+        show_mask(plt.gca(), mask, random_color=False)
+        plt.savefig(img_mask_p, bbox_inches='tight', pad_inches=0)
+        plt.close()
+
+    # inpaint the masked image
+    for idx, mask in enumerate(masks):
+        mask_p = out_dir / f"mask_{idx}.png"
+        img_inpainted_p = out_dir / f"inpainted_with_{Path(mask_p).name}"
+        img_inpainted = inpaint_img_with_lama(
+            img, mask, args.lama_config, args.lama_ckpt, device=device)
+        save_array_to_img(img_inpainted, img_inpainted_p)

replace_anything.py

@@ -0,0 +1,126 @@
+import cv2
+import sys
+import argparse
+import numpy as np
+import torch
+from pathlib import Path
+from matplotlib import pyplot as plt
+from typing import Any, Dict, List
+from sam_segment import predict_masks_with_sam
+from stable_diffusion_inpaint import replace_img_with_sd
+from utils import load_img_to_array, save_array_to_img, dilate_mask, \
+    show_mask, show_points
+
+
+def setup_args(parser):
+    parser.add_argument(
+        "--input_img", type=str, required=True,
+        help="Path to a single input img",
+    )
+    parser.add_argument(
+        "--point_coords", type=float, nargs='+', required=True,
+        help="The coordinate of the point prompt, [coord_W coord_H].",
+    )
+    parser.add_argument(
+        "--point_labels", type=int, nargs='+', required=True,
+        help="The labels of the point prompt, 1 or 0.",
+    )
+    parser.add_argument(
+        "--text_prompt", type=str, required=True,
+        help="Text prompt",
+    )
+    parser.add_argument(
+        "--dilate_kernel_size", type=int, default=None,
+        help="Dilate kernel size. Default: None",
+    )
+    parser.add_argument(
+        "--output_dir", type=str, required=True,
+        help="Output path to the directory with results.",
+    )
+    parser.add_argument(
+        "--sam_model_type", type=str,
+        default="vit_h", choices=['vit_h', 'vit_l', 'vit_b'],
+        help="The type of sam model to load. Default: 'vit_h"
+    )
+    parser.add_argument(
+        "--sam_ckpt", type=str, required=True,
+        help="The path to the SAM checkpoint to use for mask generation.",
+    )
+    parser.add_argument(
+        "--seed", type=int,
+        help="Specify seed for reproducibility.",
+    )
+    parser.add_argument(
+        "--deterministic", action="store_true",
+        help="Use deterministic algorithms for reproducibility.",
+    )
+
+
+
+if __name__ == "__main__":
+    """Example usage:
+    python replace_anything.py \
+        --input_img FA_demo/FA1_dog.png \
+        --point_coords 750 500 \
+        --point_labels 1 \
+        --text_prompt "sit on the swing" \
+        --output_dir ./results \
+        --sam_model_type "vit_h" \
+        --sam_ckpt sam_vit_h_4b8939.pth 
+    """
+    parser = argparse.ArgumentParser()
+    setup_args(parser)
+    args = parser.parse_args(sys.argv[1:])
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    img = load_img_to_array(args.input_img)
+
+    masks, _, _ = predict_masks_with_sam(
+        img,
+        [args.point_coords],
+        args.point_labels,
+        model_type=args.sam_model_type,
+        ckpt_p=args.sam_ckpt,
+        device=device,
+    )
+    masks = masks.astype(np.uint8) * 255
+
+    # dilate mask to avoid unmasked edge effect
+    if args.dilate_kernel_size is not None:
+        masks = [dilate_mask(mask, args.dilate_kernel_size) for mask in masks]
+
+    # visualize the segmentation results
+    img_stem = Path(args.input_img).stem
+    out_dir = Path(args.output_dir) / img_stem
+    out_dir.mkdir(parents=True, exist_ok=True)
+    for idx, mask in enumerate(masks):
+        # path to the results
+        mask_p = out_dir / f"mask_{idx}.png"
+        img_points_p = out_dir / f"with_points.png"
+        img_mask_p = out_dir / f"with_{Path(mask_p).name}"
+
+        # save the mask
+        save_array_to_img(mask, mask_p)
+
+        # save the pointed and masked image
+        dpi = plt.rcParams['figure.dpi']
+        height, width = img.shape[:2]
+        plt.figure(figsize=(width/dpi/0.77, height/dpi/0.77))
+        plt.imshow(img)
+        plt.axis('off')
+        show_points(plt.gca(), [args.point_coords], args.point_labels,
+                    size=(width*0.04)**2)
+        plt.savefig(img_points_p, bbox_inches='tight', pad_inches=0)
+        show_mask(plt.gca(), mask, random_color=False)
+        plt.savefig(img_mask_p, bbox_inches='tight', pad_inches=0)
+        plt.close()
+
+    # fill the masked image
+    for idx, mask in enumerate(masks):
+        if args.seed is not None:
+            torch.manual_seed(args.seed)
+        mask_p = out_dir / f"mask_{idx}.png"
+        img_replaced_p = out_dir / f"replaced_with_{Path(mask_p).name}"
+        img_replaced = replace_img_with_sd(
+            img, mask, args.text_prompt, device=device)
+        save_array_to_img(img_replaced, img_replaced_p)

requirements.txt

@@ -0,0 +1,23 @@
+torch
+torchvision
+opencv-python
+matplotlib
+tensorflow
+pyyaml
+tqdm
+numpy
+easydict
+scikit-image
+scikit-learn
+opencv-python
+joblib
+matplotlib
+pandas
+albumentations==0.5.2
+hydra-core
+pytorch-lightning
+tabulate
+kornia==0.5.0
+webdataset
+packaging
+wldhx.yadisk-direct

sam_segment.py

@@ -0,0 +1,127 @@
+import sys
+import argparse
+import numpy as np
+from pathlib import Path
+from matplotlib import pyplot as plt
+from typing import Any, Dict, List
+import torch
+
+sys.path.insert(0, str(Path(__file__).resolve().parent / "third_party" / "segment-anything"))
+
+from segment_anything import SamPredictor, sam_model_registry
+from utils import load_img_to_array, save_array_to_img, dilate_mask, \
+    show_mask, show_points
+
+
+def predict_masks_with_sam(
+        img: np.ndarray,
+        point_coords: List[List[float]],
+        point_labels: List[int],
+        model_type: str,
+        ckpt_p: str,
+        device="cuda"
+):
+    point_coords = np.array(point_coords)
+    point_labels = np.array(point_labels)
+    sam = sam_model_registry[model_type](checkpoint=ckpt_p)
+    sam.to(device=device)
+    predictor = SamPredictor(sam)
+
+    predictor.set_image(img)
+    masks, scores, logits = predictor.predict(
+        point_coords=point_coords,
+        point_labels=point_labels,
+        multimask_output=True,
+    )
+    return masks, scores, logits
+
+
+def setup_args(parser):
+    parser.add_argument(
+        "--input_img", type=str, required=True,
+        help="Path to a single input img",
+    )
+    parser.add_argument(
+        "--point_coords", type=float, nargs='+', required=True,
+        help="The coordinate of the point prompt, [coord_W coord_H].",
+    )
+    parser.add_argument(
+        "--point_labels", type=int, nargs='+', required=True,
+        help="The labels of the point prompt, 1 or 0.",
+    )
+    parser.add_argument(
+        "--dilate_kernel_size", type=int, default=None,
+        help="Dilate kernel size. Default: None",
+    )
+    parser.add_argument(
+        "--output_dir", type=str, required=True,
+        help="Output path to the directory with results.",
+    )
+    parser.add_argument(
+        "--sam_model_type", type=str,
+        default="vit_h", choices=['vit_h', 'vit_l', 'vit_b'],
+        help="The type of sam model to load. Default: 'vit_h"
+    )
+    parser.add_argument(
+        "--sam_ckpt", type=str, required=True,
+        help="The path to the SAM checkpoint to use for mask generation.",
+    )
+
+
+if __name__ == "__main__":
+    """Example usage:
+    python sam_segment.py \
+        --input_img FA_demo/FA1_dog.png \
+        --point_coords 750 500 \
+        --point_labels 1 \
+        --dilate_kernel_size 15 \
+        --output_dir ./results \
+        --sam_model_type "vit_h" \
+        --sam_ckpt sam_vit_h_4b8939.pth
+    """
+    parser = argparse.ArgumentParser()
+    setup_args(parser)
+    args = parser.parse_args(sys.argv[1:])
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    img = load_img_to_array(args.input_img)
+
+    masks, _, _ = predict_masks_with_sam(
+        img,
+        [args.point_coords],
+        args.point_labels,
+        model_type=args.sam_model_type,
+        ckpt_p=args.sam_ckpt,
+        device=device,
+    )
+    masks = masks.astype(np.uint8) * 255
+
+    # dilate mask to avoid unmasked edge effect
+    if args.dilate_kernel_size is not None:
+        masks = [dilate_mask(mask, args.dilate_kernel_size) for mask in masks]
+
+    # visualize the segmentation results
+    img_stem = Path(args.input_img).stem
+    out_dir = Path(args.output_dir) / img_stem
+    out_dir.mkdir(parents=True, exist_ok=True)
+    for idx, mask in enumerate(masks):
+        # path to the results
+        mask_p = out_dir / f"mask_{idx}.png"
+        img_points_p = out_dir / f"with_points.png"
+        img_mask_p = out_dir / f"with_{Path(mask_p).name}"
+
+        # save the mask
+        save_array_to_img(mask, mask_p)
+
+        # save the pointed and masked image
+        dpi = plt.rcParams['figure.dpi']
+        height, width = img.shape[:2]
+        plt.figure(figsize=(width/dpi/0.77, height/dpi/0.77))
+        plt.imshow(img)
+        plt.axis('off')
+        show_points(plt.gca(), [args.point_coords], args.point_labels,
+                    size=(width*0.04)**2)
+        plt.savefig(img_points_p, bbox_inches='tight', pad_inches=0)
+        show_mask(plt.gca(), mask, random_color=False)
+        plt.savefig(img_mask_p, bbox_inches='tight', pad_inches=0)
+        plt.close()

stable_diffusion_inpaint.py

@@ -0,0 +1,117 @@
+import os
+import sys
+import glob
+import argparse
+import torch
+import numpy as np
+import PIL.Image as Image
+from pathlib import Path
+from diffusers import StableDiffusionInpaintPipeline
+from utils.mask_processing import crop_for_filling_pre, crop_for_filling_post
+from utils.crop_for_replacing import recover_size, resize_and_pad
+from utils import load_img_to_array, save_array_to_img
+
+
+def fill_img_with_sd(
+        img: np.ndarray,
+        mask: np.ndarray,
+        text_prompt: str,
+        device="cuda"
+):
+    pipe = StableDiffusionInpaintPipeline.from_pretrained(
+        "stabilityai/stable-diffusion-2-inpainting",
+        torch_dtype=torch.float32,
+    ).to(device)
+    img_crop, mask_crop = crop_for_filling_pre(img, mask)
+    img_crop_filled = pipe(
+        prompt=text_prompt,
+        image=Image.fromarray(img_crop),
+        mask_image=Image.fromarray(mask_crop)
+    ).images[0]
+    img_filled = crop_for_filling_post(img, mask, np.array(img_crop_filled))
+    return img_filled
+
+
+def replace_img_with_sd(
+        img: np.ndarray,
+        mask: np.ndarray,
+        text_prompt: str,
+        step: int = 50,
+        device="cuda"
+):
+    pipe = StableDiffusionInpaintPipeline.from_pretrained(
+        "stabilityai/stable-diffusion-2-inpainting",
+        torch_dtype=torch.float32,
+    ).to(device)
+    img_padded, mask_padded, padding_factors = resize_and_pad(img, mask)
+    img_padded = pipe(
+        prompt=text_prompt,
+        image=Image.fromarray(img_padded),
+        mask_image=Image.fromarray(255 - mask_padded),
+        num_inference_steps=step,
+    ).images[0]
+    height, width, _ = img.shape
+    img_resized, mask_resized = recover_size(
+        np.array(img_padded), mask_padded, (height, width), padding_factors)
+    mask_resized = np.expand_dims(mask_resized, -1) / 255
+    img_resized = img_resized * (1-mask_resized) + img * mask_resized
+    return img_resized
+
+
+def setup_args(parser):
+    parser.add_argument(
+        "--input_img", type=str, required=True,
+        help="Path to a single input img",
+    )
+    parser.add_argument(
+        "--text_prompt", type=str, required=True,
+        help="Text prompt",
+    )
+    parser.add_argument(
+        "--input_mask_glob", type=str, required=True,
+        help="Glob to input masks",
+    )
+    parser.add_argument(
+        "--output_dir", type=str, required=True,
+        help="Output path to the directory with results.",
+    )
+    parser.add_argument(
+        "--seed", type=int,
+        help="Specify seed for reproducibility.",
+    )
+    parser.add_argument(
+        "--deterministic", action="store_true",
+        help="Use deterministic algorithms for reproducibility.",
+    )
+
+if __name__ == "__main__":
+    """Example usage:
+    python lama_inpaint.py \
+        --input_img FA_demo/FA1_dog.png \
+        --input_mask_glob "results/FA1_dog/mask*.png" \
+        --text_prompt "a teddy bear on a bench" \
+        --output_dir results 
+    """
+    parser = argparse.ArgumentParser()
+    setup_args(parser)
+    args = parser.parse_args(sys.argv[1:])
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    if args.deterministic:
+        os.environ["CUBLAS_WORKSPACE_CONFIG"] = ":4096:8"
+        torch.use_deterministic_algorithms(True)
+
+    img_stem = Path(args.input_img).stem
+    mask_ps = sorted(glob.glob(args.input_mask_glob))
+    out_dir = Path(args.output_dir) / img_stem
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    img = load_img_to_array(args.input_img)
+    for mask_p in mask_ps:
+        if args.seed is not None:
+            torch.manual_seed(args.seed)
+        mask = load_img_to_array(mask_p)
+        img_filled_p = out_dir / f"filled_with_{Path(mask_p).name}"
+        img_filled = fill_img_with_sd(
+            img, mask, args.text_prompt, device=device)
+        save_array_to_img(img_filled, img_filled_p)

third_party/.gitkeep

@@ -0,0 +1 @@
+# Put third party packages into this directory

third_party/lama/.gitignore

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbe005fc6c62133aee2511011dad6e169d913bba51490d29d36659566c7d84e9
+size 1865

third_party/lama/LICENSE

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ceeeac5a802e86c413c22b16cce8e9a22027b0250c97e6f8ac97c14cf0542c0
+size 11348

third_party/lama/README.md

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54b097c25246f2193703790ca10f3da7a134ca9cd6fee6f7f14d1e03c016df3a
+size 16575

third_party/lama/bin/analyze_errors.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9019332b8309090906f49ac63fe6a6e5f1a5a03e2eedf3e56c8be6cfd6fb2dcc
+size 17698

third_party/lama/bin/blur_predicts.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed0863c166ba510175e89a373f7b950550e8008ed4b328e6b7d2c53e7237889a
+size 2191

third_party/lama/bin/calc_dataset_stats.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53e279cba5626e01ffaa7984b775603f64cdf149a72fa16a0f58000a3bcbc4ec
+size 3627

third_party/lama/bin/debug/analyze_overlapping_masks.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c368f947b87d27ecda000ba96d2dd81437aa24215064f1cae1df92c86f0f1c52
+size 1132

third_party/lama/bin/evaluate_predicts.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67458a187713e1455dd1ff013bd678d608979ae49f637a22f24ee92d40de0e8b
+size 3530

third_party/lama/bin/evaluator_example.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c88b5a0d15bfa6e81b60c963aa8f32ff752b853ae3cf38966d24cccdfc8b931
+size 2359

third_party/lama/bin/extract_masks.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e90a3197316ee22fd33b01ce1df8b8e21be39288e1669ec1c9f4d7e6e0fcd19
+size 1627

third_party/lama/bin/filter_sharded_dataset.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48c820e8ec725ffbc1b625d6924dab43c5134f955e0a32bc3ee756f437e6c2f1
+size 2346

third_party/lama/bin/gen_debug_mask_dataset.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea1b7f9144b3716ebe676bb89efd9ae052e4831670ba65c3b1f05ef770da5e76
+size 1909

third_party/lama/bin/gen_mask_dataset.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f0ab3421b1b5f69b00e2e8faec6ac6ef7f980dc8ce8aed340bd99e3e5bd5a1b
+size 5609

third_party/lama/bin/gen_mask_dataset_hydra.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d732574c27a74fa64c66a1afea6497e4fd87103d7da3d58e3dc83f7bb59ad46e
+size 5340

third_party/lama/bin/gen_outpainting_dataset.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7457d08e1f495c3bee3de2413fdb735f252d3fe342b5e54e5372af73b1598211
+size 3226

third_party/lama/bin/make_checkpoint.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d86903c380809c4c266a0ea39cae4d2fb4aa4fcae9ca51491053ac28a72c2edf
+size 3097

third_party/lama/bin/mask_example.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3d14ee869193ebb1b5ac68ea92c9506e7f9bba280ca9c4581fb170cd6ccd789
+size 466

third_party/lama/bin/paper_runfiles/blur_tests.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c8ec2117f3e600321150ed0f24d385c71edb14f2c221a2c32afe6d61a833cfc
+size 1491

third_party/lama/bin/paper_runfiles/env.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:669a5d8d21c3c4400225ef11a4d424d79de578cd5fa923ef55fb930e743a7094
+size 170

third_party/lama/bin/paper_runfiles/find_best_checkpoint.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42d198b58e58ec822e1aac66d66251b87c519f8b9ec178a1c0f810f0891841ae
+size 2075

third_party/lama/bin/paper_runfiles/generate_test_celeba-hq.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecde53349bd3df42ce1e2659578ab493f78a521760168c1458c746b06c304fe9
+size 603

third_party/lama/bin/paper_runfiles/generate_test_ffhq.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27abaa6c56ac539bea34bac171d31558826399e1ea2e69742ac0bb9f93364413
+size 580

third_party/lama/bin/paper_runfiles/generate_test_paris.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b48c06f2d0197d5b52fd6dbaa3c3b24ec41d19e8ae485e920eab1f85b9e2d0c
+size 593

third_party/lama/bin/paper_runfiles/generate_test_paris_256.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:174c38fc6663c4bb6eb7b259a81ec6e4b1d16df0ae40cd4329b2fb4656306a74
+size 598

third_party/lama/bin/paper_runfiles/generate_val_test.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13ea073dbdec3dcab2af58f39099f1e23f7a93e7f445ba00d3091070ab1cd329
+size 1045

third_party/lama/bin/paper_runfiles/predict_inner_features.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eebdbcad0999666cc465eea6cf5154bd90a933a044da95888de6e6f1fdf5b15a
+size 841

third_party/lama/bin/paper_runfiles/update_test_data_stats.sh

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:686112f14750df31bb8dec7d603088b40a2b6b2083df6688817640c7d26d788c
+size 927

third_party/lama/bin/predict.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d7cff7b5f498e895830f97fc076f33ad61bf62c5a117f2028932b267e9d13f7
+size 4092

third_party/lama/bin/predict_inner_features.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2754aedde8a69795dca8cb2ee2919ba6237397b1d0501061f684e347d35128ac
+size 4938

third_party/lama/bin/report_from_tb.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c110718a1f0c37ed143eeac317a4bd3522350060e178ea655fd5d0bd32804dfc
+size 2588

third_party/lama/bin/sample_from_dataset.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb5507435eb50c10e50f96707e77f5109b8c8f73f14c520a24b321b24b6fc642
+size 3044

third_party/lama/bin/side_by_side.py

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9eaec563dbc98868725083fd560f729e3e50dc263b71a6ff1bbfd9f627df0b5d
+size 2826