This is the official repository of the paper "Back-Hand-Pose: 3D Hand Pose Estimation for Wrist-worn Camera via Dorsum Deformation Network" which deal with estimating human hand pose from a wrist-worn camera. The deformation of the back of hand is analyzed and mapped to a special joint angle-based 3D hand representation, and the joint angle is then reconstructed to 3D Hand Pose by Unity and IK.

This demonstration here provide the several contents:

1. The data preprocessing network for generating masked dorsal hand.
2. Estimating real-time joint angle from masked dorsal hand.

To be updated

To be updated

To be updated

This is the official repository of the paper "Back-Hand-Pose: 3D Hand Pose Estimation for Wrist-wornDevices via Dorsum Deformation Analysis" (UIST 2020).

If you refer to our paper or use our code, please also cite this paper and the previous work:

@inproceedings{10.1145/3379337.3415897,
author = {Wu, Erwin and Yuan, Ye and Yeo, Hui-Shyong and Quigley, Aaron and Koike, Hideki and Kitani, Kris M.},
title = {Back-Hand-Pose: 3D Hand Pose Estimation for a Wrist-Worn Camera via Dorsum Deformation Network},
year = {2020},
isbn = {9781450375146},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3379337.3415897},
doi = {10.1145/3379337.3415897},
booktitle = {Proceedings of the 33rd Annual ACM Symposium on User Interface Software and Technology},
pages = {1147–1160},
numpages = {14},
keywords = {3d hand pose estimation, dorsal hand, wrist-worn devices},
location = {Virtual Event, USA},
series = {UIST '20}
}

The code, software, and data in this repository is only available for non-commercial research use. Please see the license for further details.