Add source code

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.

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 +1,134 @@
+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 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)

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)

sam_segment.py

@@ -0,0 +1,125 @@
+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
+
+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)

utils/__init__.py

@@ -0,0 +1 @@
+from .utils import *

utils/crop_for_replacing.py

@@ -0,0 +1,101 @@
+import cv2
+import numpy as np
+from typing import Tuple
+
+def resize_and_pad(image: np.ndarray, mask: np.ndarray, target_size: int = 512) -> Tuple[np.ndarray, np.ndarray]:
+    """
+    Resizes an image and its corresponding mask to have the longer side equal to `target_size` and pads them to make them
+    both have the same size. The resulting image and mask have dimensions (target_size, target_size).
+
+    Args:
+        image: A numpy array representing the image to resize and pad.
+        mask: A numpy array representing the mask to resize and pad.
+        target_size: An integer specifying the desired size of the longer side after resizing.
+
+    Returns:
+        A tuple containing two numpy arrays - the resized and padded image and the resized and padded mask.
+    """
+    height, width, _ = image.shape
+    max_dim = max(height, width)
+    scale = target_size / max_dim
+    new_height = int(height * scale)
+    new_width = int(width * scale)
+    image_resized = cv2.resize(image, (new_width, new_height), interpolation=cv2.INTER_LINEAR)
+    mask_resized = cv2.resize(mask, (new_width, new_height), interpolation=cv2.INTER_LINEAR)
+    pad_height = target_size - new_height
+    pad_width = target_size - new_width
+    top_pad = pad_height // 2
+    bottom_pad = pad_height - top_pad
+    left_pad = pad_width // 2
+    right_pad = pad_width - left_pad
+    image_padded = np.pad(image_resized, ((top_pad, bottom_pad), (left_pad, right_pad), (0, 0)), mode='constant')
+    mask_padded = np.pad(mask_resized, ((top_pad, bottom_pad), (left_pad, right_pad)), mode='constant')
+    return image_padded, mask_padded, (top_pad, bottom_pad, left_pad, right_pad)
+
+def recover_size(image_padded: np.ndarray, mask_padded: np.ndarray, orig_size: Tuple[int, int], 
+                 padding_factors: Tuple[int, int, int, int]) -> Tuple[np.ndarray, np.ndarray]:
+    """
+    Resizes a padded and resized image and mask to the original size.
+
+    Args:
+        image_padded: A numpy array representing the padded and resized image.
+        mask_padded: A numpy array representing the padded and resized mask.
+        orig_size: A tuple containing two integers - the original height and width of the image before resizing and padding.
+
+    Returns:
+        A tuple containing two numpy arrays - the recovered image and the recovered mask with dimensions `orig_size`.
+    """
+    h,w,c = image_padded.shape
+    top_pad, bottom_pad, left_pad, right_pad = padding_factors
+    image = image_padded[top_pad:h-bottom_pad, left_pad:w-right_pad, :]
+    mask = mask_padded[top_pad:h-bottom_pad, left_pad:w-right_pad]
+    image_resized = cv2.resize(image, orig_size[::-1], interpolation=cv2.INTER_LINEAR)
+    mask_resized = cv2.resize(mask, orig_size[::-1], interpolation=cv2.INTER_LINEAR)
+    return image_resized, mask_resized
+
+
+
+
+if __name__ == '__main__':
+
+    # image = cv2.imread('example/boat.jpg')
+    # mask = cv2.imread('example/boat_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    # image = cv2.imread('example/groceries.jpg')
+    # mask = cv2.imread('example/groceries_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    # image = cv2.imread('example/bridge.jpg')
+    # mask = cv2.imread('example/bridge_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    # image = cv2.imread('example/person_umbrella.jpg')
+    # mask = cv2.imread('example/person_umbrella_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    # image = cv2.imread('example/hippopotamus.jpg')
+    # mask = cv2.imread('example/hippopotamus_mask_1.png', cv2.IMREAD_GRAYSCALE)
+    image = cv2.imread('/data1/yutao/projects/IAM/Inpaint-Anything/example/fill-anything/sample5.jpeg')
+    mask = cv2.imread('/data1/yutao/projects/IAM/Inpaint-Anything/example/fill-anything/sample5/mask.png', cv2.IMREAD_GRAYSCALE)
+    print(image.shape)
+    print(mask.shape)
+    cv2.imwrite('original_image.jpg', image)
+    cv2.imwrite('original_mask.jpg', mask)
+    image_padded, mask_padded, padding_factors = resize_and_pad(image, mask)
+    cv2.imwrite('padded_image.png', image_padded)
+    cv2.imwrite('padded_mask.png', mask_padded)
+    print(image_padded.shape, mask_padded.shape, padding_factors)
+
+    # ^ ------------------------------------------------------------------------------------
+    # ^ Please conduct inpainting or filling here on the cropped image with the cropped mask
+    # ^ ------------------------------------------------------------------------------------
+
+    # resize and pad the image and mask
+
+    # perform some operation on the 512x512 image and mask
+    # ...
+
+    # recover the image and mask to the original size
+    height, width, _ = image.shape
+    image_resized, mask_resized = recover_size(image_padded, mask_padded, (height, width), padding_factors)
+
+    # save the resized and recovered image and mask
+    cv2.imwrite('resized_and_padded_image.png', image_padded)
+    cv2.imwrite('resized_and_padded_mask.png', mask_padded)
+    cv2.imwrite('recovered_image.png', image_resized)
+    cv2.imwrite('recovered_mask.png', mask_resized)
+
+        

utils/get_point_coor.py

@@ -0,0 +1,12 @@
+import cv2
+
+def click_event(event, x, y, flags, param):
+    if event == cv2.EVENT_LBUTTONDOWN:
+        print("Point coordinates ({}, {})".format(x, y))
+img = cv2.imread("./example/remove-anything/dog.jpg")
+
+cv2.imshow("Image", img)
+cv2.setMouseCallback("Image", click_event)
+cv2.waitKey(0)
+
+cv2.destroyAllWindows()

utils/mask_processing.py

@@ -0,0 +1,160 @@
+import cv2
+from matplotlib import pyplot as plt
+import PIL.Image as Image
+import numpy as np
+
+
+def crop_for_filling_pre(image: np.array, mask: np.array, crop_size: int = 512):
+    # Calculate the aspect ratio of the image
+    height, width = image.shape[:2]
+    aspect_ratio = float(width) / float(height)
+
+    # If the shorter side is less than 512, resize the image proportionally
+    if min(height, width) < crop_size:
+        if height < width:
+            new_height = crop_size
+            new_width = int(new_height * aspect_ratio)
+        else:
+            new_width = crop_size
+            new_height = int(new_width / aspect_ratio)
+
+        image = cv2.resize(image, (new_width, new_height))
+        mask = cv2.resize(mask, (new_width, new_height))
+
+    # Find the bounding box of the mask
+    x, y, w, h = cv2.boundingRect(mask)
+
+    # Update the height and width of the resized image
+    height, width = image.shape[:2]
+
+    # # If the 512x512 square cannot cover the entire mask, resize the image accordingly
+    if w > crop_size or h > crop_size:
+        # padding to square at first
+        if height < width:
+            padding = width - height
+            image = np.pad(image, ((padding // 2, padding - padding // 2), (0, 0), (0, 0)), 'constant')
+            mask = np.pad(mask, ((padding // 2, padding - padding // 2), (0, 0)), 'constant')
+        else:
+            padding = height - width
+            image = np.pad(image, ((0, 0), (padding // 2, padding - padding // 2), (0, 0)), 'constant')
+            mask = np.pad(mask, ((0, 0), (padding // 2, padding - padding // 2)), 'constant')
+
+        resize_factor = crop_size / max(w, h)
+        image = cv2.resize(image, (0, 0), fx=resize_factor, fy=resize_factor)
+        mask = cv2.resize(mask, (0, 0), fx=resize_factor, fy=resize_factor)
+        x, y, w, h = cv2.boundingRect(mask)
+
+    # Calculate the crop coordinates
+    crop_x = min(max(x + w // 2 - crop_size // 2, 0), width - crop_size)
+    crop_y = min(max(y + h // 2 - crop_size // 2, 0), height - crop_size)
+
+    # Crop the image
+    cropped_image = image[crop_y:crop_y + crop_size, crop_x:crop_x + crop_size]
+    cropped_mask = mask[crop_y:crop_y + crop_size, crop_x:crop_x + crop_size]
+
+    return cropped_image, cropped_mask
+    
+    
+def crop_for_filling_post(
+        image: np.array,
+        mask: np.array,
+        filled_image: np.array, 
+        crop_size: int = 512,
+        ):
+    image_copy = image.copy()
+    mask_copy = mask.copy()
+    # Calculate the aspect ratio of the image
+    height, width = image.shape[:2]
+    height_ori, width_ori = height, width
+    aspect_ratio = float(width) / float(height)
+
+    # If the shorter side is less than 512, resize the image proportionally
+    if min(height, width) < crop_size:
+        if height < width:
+            new_height = crop_size
+            new_width = int(new_height * aspect_ratio)
+        else:
+            new_width = crop_size
+            new_height = int(new_width / aspect_ratio)
+
+        image = cv2.resize(image, (new_width, new_height))
+        mask = cv2.resize(mask, (new_width, new_height))
+
+    # Find the bounding box of the mask
+    x, y, w, h = cv2.boundingRect(mask)
+
+    # Update the height and width of the resized image
+    height, width = image.shape[:2]
+
+    # # If the 512x512 square cannot cover the entire mask, resize the image accordingly
+    if w > crop_size or h > crop_size:
+        flag_padding = True
+        # padding to square at first
+        if height < width:
+            padding = width - height
+            image = np.pad(image, ((padding // 2, padding - padding // 2), (0, 0), (0, 0)), 'constant')
+            mask = np.pad(mask, ((padding // 2, padding - padding // 2), (0, 0)), 'constant')
+            padding_side = 'h'
+        else:
+            padding = height - width
+            image = np.pad(image, ((0, 0), (padding // 2, padding - padding // 2), (0, 0)), 'constant')
+            mask = np.pad(mask, ((0, 0), (padding // 2, padding - padding // 2)), 'constant')
+            padding_side = 'w'
+
+        resize_factor = crop_size / max(w, h)
+        image = cv2.resize(image, (0, 0), fx=resize_factor, fy=resize_factor)
+        mask = cv2.resize(mask, (0, 0), fx=resize_factor, fy=resize_factor)
+        x, y, w, h = cv2.boundingRect(mask)
+    else:
+        flag_padding = False
+
+    # Calculate the crop coordinates
+    crop_x = min(max(x + w // 2 - crop_size // 2, 0), width - crop_size)
+    crop_y = min(max(y + h // 2 - crop_size // 2, 0), height - crop_size)
+
+    # Fill the image
+    image[crop_y:crop_y + crop_size, crop_x:crop_x + crop_size] = filled_image
+    if flag_padding:
+        image = cv2.resize(image, (0, 0), fx=1/resize_factor, fy=1/resize_factor)
+        if padding_side == 'h':
+            image = image[padding // 2:padding // 2 + height_ori, :]
+        else:
+            image = image[:, padding // 2:padding // 2 + width_ori]
+
+    image = cv2.resize(image, (width_ori, height_ori))
+
+    image_copy[mask_copy==255] = image[mask_copy==255]
+    return image_copy
+
+
+if __name__ == '__main__':
+
+    # image = cv2.imread('example/boat.jpg')
+    # mask = cv2.imread('example/boat_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    image = cv2.imread('./example/groceries.jpg')
+    mask = cv2.imread('example/groceries_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    # image = cv2.imread('example/bridge.jpg')
+    # mask = cv2.imread('example/bridge_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    # image = cv2.imread('example/person_umbrella.jpg')
+    # mask = cv2.imread('example/person_umbrella_mask_2.png', cv2.IMREAD_GRAYSCALE)
+    # image = cv2.imread('example/hippopotamus.jpg')
+    # mask = cv2.imread('example/hippopotamus_mask_1.png', cv2.IMREAD_GRAYSCALE)
+
+    cropped_image, cropped_mask = crop_for_filling_pre(image, mask)
+    # ^ ------------------------------------------------------------------------------------
+    # ^ Please conduct inpainting or filling here on the cropped image with the cropped mask
+    # ^ ------------------------------------------------------------------------------------
+
+    # e.g.
+    # cropped_image[cropped_mask==255] = 0
+    cv2.imwrite('cropped_image.jpg', cropped_image)
+    cv2.imwrite('cropped_mask.jpg', cropped_mask)
+    print(cropped_image.shape)
+    print(cropped_mask.shape)
+
+    image = crop_for_filling_post(image, mask, cropped_image)
+    cv2.imwrite('filled_image.jpg', image)
+    print(image.shape)
+
+
+    

utils/paste_object.py

@@ -0,0 +1,50 @@
+import cv2
+import numpy as np
+
+def paste_object(source, source_mask, target, target_coords, resize_scale=1):
+    assert target_coords[0] < target.shape[1] and target_coords[1] < target.shape[0]
+    # Find the bounding box of the source_mask
+    x, y, w, h = cv2.boundingRect(source_mask)
+    assert h < source.shape[0] and w < source.shape[1]
+    obj = source[y:y+h, x:x+w]
+    obj_msk = source_mask[y:y+h, x:x+w]
+    if resize_scale != 1:
+        obj = cv2.resize(obj, (0,0), fx=resize_scale, fy=resize_scale)
+        obj_msk = cv2.resize(obj_msk, (0,0), fx=resize_scale, fy=resize_scale)
+        _, _, w, h = cv2.boundingRect(obj_msk)
+
+    xt = max(0, target_coords[0]-w//2)
+    yt = max(0, target_coords[1]-h//2)
+    if target_coords[0]-w//2 < 0:
+        obj = obj[:, w//2-target_coords[0]:]
+        obj_msk = obj_msk[:, w//2-target_coords[0]:]
+    if target_coords[0]+w//2 > target.shape[1]:
+        obj = obj[:, :target.shape[1]-target_coords[0]+w//2]
+        obj_msk = obj_msk[:, :target.shape[1]-target_coords[0]+w//2]
+    if target_coords[1]-h//2 < 0:
+        obj = obj[h//2-target_coords[1]:, :]
+        obj_msk = obj_msk[h//2-target_coords[1]:, :]
+    if target_coords[1]+h//2 > target.shape[0]:
+        obj = obj[:target.shape[0]-target_coords[1]+h//2, :]
+        obj_msk = obj_msk[:target.shape[0]-target_coords[1]+h//2, :]
+    _, _, w, h = cv2.boundingRect(obj_msk)
+
+    target[yt:yt+h, xt:xt+w][obj_msk==255] = obj[obj_msk==255]
+    target_mask = np.zeros_like(target)
+    target_mask = cv2.cvtColor(target_mask, cv2.COLOR_BGR2GRAY)
+    target_mask[yt:yt+h, xt:xt+w][obj_msk==255] = 255
+    
+    return target, target_mask
+
+if __name__ == '__main__':
+    source = cv2.imread('example/boat.jpg')
+    source_mask = cv2.imread('example/boat_mask_1.png', 0)
+    target = cv2.imread('example/hippopotamus.jpg')
+    print(source.shape, source_mask.shape, target.shape)
+    
+    target_coords = (700, 400)  # (x, y)
+    resize_scale = 1
+    target, target_mask = paste_object(source, source_mask, target, target_coords, resize_scale)
+    cv2.imwrite('target_pasted.png', target)
+    cv2.imwrite('target_mask.png', target_mask)
+    print(target.shape, target_mask.shape)

utils/utils.py

@@ -0,0 +1,53 @@
+import cv2
+import numpy as np
+from PIL import Image
+from typing import Any, Dict, List
+
+
+def load_img_to_array(img_p):
+    return np.array(Image.open(img_p))
+
+
+def save_array_to_img(img_arr, img_p):
+    Image.fromarray(img_arr.astype(np.uint8)).save(img_p)
+
+
+def dilate_mask(mask, dilate_factor=15):
+    mask = mask.astype(np.uint8)
+    mask = cv2.dilate(
+        mask,
+        np.ones((dilate_factor, dilate_factor), np.uint8),
+        iterations=1
+    )
+    return mask
+
+def erode_mask(mask, dilate_factor=15):
+    mask = mask.astype(np.uint8)
+    mask = cv2.erode(
+        mask,
+        np.ones((dilate_factor, dilate_factor), np.uint8),
+        iterations=1
+    )
+    return mask
+
+def show_mask(ax, mask: np.ndarray, random_color=False):
+    mask = mask.astype(np.uint8)
+    if np.max(mask) == 255:
+        mask = mask / 255
+    if random_color:
+        color = np.concatenate([np.random.random(3), np.array([0.6])], axis=0)
+    else:
+        color = np.array([30 / 255, 144 / 255, 255 / 255, 0.6])
+    h, w = mask.shape[-2:]
+    mask_img = mask.reshape(h, w, 1) * color.reshape(1, 1, -1)
+    ax.imshow(mask_img)
+
+
+def show_points(ax, coords: List[List[float]], labels: List[int], size=375):
+    coords = np.array(coords)
+    labels = np.array(labels)
+    color_table = {0: 'red', 1: 'green'}
+    for label_value, color in color_table.items():
+        points = coords[labels == label_value]
+        ax.scatter(points[:, 0], points[:, 1], color=color, marker='*',
+                   s=size, edgecolor='white', linewidth=1.25)

utils/visual_mask_on_img.py

@@ -0,0 +1,62 @@
+import cv2
+import sys
+import argparse
+import numpy as np
+from PIL import Image
+from pathlib import Path
+from matplotlib import pyplot as plt
+from typing import Any, Dict, List
+import glob
+
+from utils import load_img_to_array, show_mask
+
+
+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.",
+    )
+
+if __name__ == "__main__":
+    """Example usage:
+    python visual_mask_on_img.py \
+        --input_img FA_demo/FA1_dog.png \
+        --input_mask_glob "results/FA1_dog/mask*.png" \
+        --output_dir results
+    """
+    parser = argparse.ArgumentParser()
+    setup_args(parser)
+    args = parser.parse_args(sys.argv[1:])
+
+    img = load_img_to_array(args.input_img)
+    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)
+
+    for mask_p in mask_ps:
+        mask = load_img_to_array(mask_p)
+        mask = mask.astype(np.uint8)
+
+        # path to the results
+        img_mask_p = out_dir / f"with_{Path(mask_p).name}"
+
+        # save the 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_mask(plt.gca(), mask, random_color=False)
+        plt.savefig(img_mask_p, bbox_inches='tight', pad_inches=0)
+        plt.close()