fashn-ai/DWPose

DWPose - ONNX Weights

This repository hosts ONNX weights for DWPose, a state-of-the-art whole-body pose estimation model. These weights are provided for convenience to support the FASHN Virtual Try-On pipeline.

Attribution

This is NOT an original work by FASHN AI. We are hosting these weights solely for ease of integration with our virtual try-on pipeline.

Original Authors

DWPose was created by researchers at IDEA-Research:

• Zhendong Yang
• Ailing Zeng
• Chun Yuan
• Yu Li

Original Repository

GitHub: https://github.com/IDEA-Research/DWPose

Please visit the original repository for:

• Full documentation
• Training code
• Additional model variants
• Issues and discussions

Paper

"Effective Whole-body Pose Estimation with Two-stages Distillation"

Published at ICCV 2023 (CV4Metaverse Workshop)

• arXiv: https://arxiv.org/abs/2307.15880
• ICCV Open Access: https://openaccess.thecvf.com/content/ICCV2023W/CV4Metaverse/html/Yang_Effective_Whole-Body_Pose_Estimation_with_Two-Stages_Distillation_ICCVW_2023_paper.html

Citation

If you use DWPose in your research, please cite the original authors:

@inproceedings{yang2023effective,
  title={Effective Whole-body Pose Estimation with Two-stages Distillation},
  author={Yang, Zhendong and Zeng, Ailing and Yuan, Chun and Li, Yu},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops},
  pages={4210--4220},
  year={2023}
}

Model Files

Acknowledgments

DWPose builds upon the following projects:

• MMPose - OpenMMLab pose estimation toolbox
• RTMPose - Real-Time Multi-Person Pose Estimation
• YOLOX - High-performance object detector

License

DWPose is released under the Apache 2.0 License, which permits commercial use with proper attribution.

Note: FASHN AI does not claim any ownership of DWPose. We are grateful to the original authors for making their work available under an open-source license. For any questions about the model itself, please refer to the original repository.