Accepted in Pattern Recognition (PR) 2023.
mkdir pose2pose
cd pose2pose
mkdir -p datasets/DeepFashion
mkdir -p output/DeepFashion/ckpt/pretrained
git clone https://github.com/prasunroy/pose-transfer.git
cd pose-transfer
pip install -r requirements.txtfrom api import Pose2Pose
from PIL import Image
p2p = Pose2Pose(pretrained=True)
condition = Image.open('./api-test/condition.jpg')
reference = Image.open('./api-test/target_pose_reference.jpg')
generated = p2p.transfer_as(condition, reference)
generated.show()- Download dataset files from Google Drive and extract into
datasets/DeepFashiondirectory. - Download pretrained checkpoints from Google Drive into
output/DeepFashion/ckpt/pretraineddirectory.
pose2pose
│
├───datasets
│ └───DeepFashion
│ ├───img
│ ├───test_pose_maps
│ ├───train_pose_maps
│ ├───test_img_keypoints.csv
│ ├───test_img_list.csv
│ ├───test_img_pairs.csv
│ ├───train_img_keypoints.csv
│ ├───train_img_list.csv
│ └───train_img_pairs.csv
├───output
│ └───DeepFashion
│ └───ckpt
│ └───pretrained
│ ├───netD_257500.pth
│ ├───netD_260500.pth
│ ├───netG_257500.pth
│ └───netG_260500.pth
└───pose-transfer
The precomputed keypoints and posemaps are estimated using the provided utility scripts in pose-transfer/utils.
In the paper, all qualitative results are generated using the pretrained checkpoint at iteration 260500 and all quantitative evaluations are performed using the pretrained checkpoint at iteration 257500.
@article{roy2022multi,
title = {Multi-scale Attention Guided Pose Transfer},
author = {Roy, Prasun and Bhattacharya, Saumik and Ghosh, Subhankar and Pal, Umapada},
journal = {Pattern Recognition},
volume = {137},
pages = {109315},
year = {2023},
issn = {0031-3203},
doi = {https://doi.org/10.1016/j.patcog.2023.109315}
}
Copyright 2023 by the authors
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.
The pretrained models are released under Creative Commons Attribution 4.0 International (CC BY 4.0) license.
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot]")
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent