Minimal-Hand

This is a unofficial implementation of Minimal-Hand based on PyTorch(CVPR2020).

This project provides the core components for hand motion capture:

  • estimating joint locations from a monocular RGB image (DetNet)
  • estimating joint rotations from locations (IKNet)

DO

  • [x] Inference code like offical one
  • [ ] Training code

Quick Start

Download

First, you need download detnet and iknet transfered from tensorflow version. Note that there is a bit of precision error when I transfer tensorflow module to pytorch:

And then, you need mkdir weights into which you put the weights downloaded from the above websites.

Inference

  python app.py

Note that you could choose your video parameters by modify the following code in app.py

  live_application(video_stream("../test_video/img_3/"))---> live_application(video_stream("../your_image_files_path"))

Visualization

000033

000043

Citation

This is the unofficial implementation of the paper "Monocular Real-time Hand Shape and Motion Capture using Multi-modal Data" (CVPR 2020).

If you find the project helpful, please consider citing them:

@inproceedings{zhou2020monocular,
  title={Monocular Real-time Hand Shape and Motion Capture using Multi-modal Data},
  author={Zhou, Yuxiao and Habermann, Marc and Xu, Weipeng and Habibie, Ikhsanul and Theobalt, Christian and Xu, Feng},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={5346--5355},
  year={2020}
}

GitHub