MMSkeleton is an open source toolbox for skeleton-based human understanding. It is a part of the open-mmlab project in the charge of Multimedia Laboratory, CUHK. MMSkeleton is developed on our research project ST-GCN.
- [2019-08-29] MMSkeleton v0.5 is released.
- [2019-09-23] Add video-based pose estimation demo.
-
High extensibility
MMSkeleton provides a flexible framework for organizing codes and projects systematically, with the ability to extend to various tasks and scale up to complex deep models.
-
Multiple tasks
MMSkeleton addresses to multiple tasks in human understanding, including but not limited to:
- skeleton-based action recognition: [ST-GCN]
- 2D pose estimation: [START_POSE_ESTIMATION.md]
- skeleton-based action generation
- 3D pose estimation
- pose tracking
Please see GETTING_STARTED.md and START_RECOGNITION.md for more details of MMSkeleton.
The project is release under the Apache 2.0 license.
We appreciate all contributions to improve MMSkeleton. Please refer to CONTRIBUTING.md for the contributing guideline.
Please cite the following paper if you use this repository in your reseach.
@inproceedings{stgcn2018aaai,
title = {Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition},
author = {Sijie Yan and Yuanjun Xiong and Dahua Lin},
booktitle = {AAAI},
year = {2018},
}
For any question, feel free to contact
Sijie Yan : [email protected]
Yuanjun Xiong : [email protected]