kenziyuliu/MS-G3D
PyTorch implementation of graph convolution networks for recognizing human actions from 3D skeleton sequences.
Not currently ranked — collecting fresh signals.
star history
This repository implements MS-G3D, a CVPR 2020 Oral paper on skeleton-based action recognition using graph convolutional networks. The model disentangles and unifies graph convolutions to better capture spatial-temporal patterns in human skeleton data. It supports training on datasets like NTU RGB+D 60/120 and Kinetics 400, and includes pretrained model weights for evaluation.
Frequently asked
- What is kenziyuliu/MS-G3D?
- PyTorch implementation of graph convolution networks for recognizing human actions from 3D skeleton sequences.
- Is MS-G3D open source?
- Yes — kenziyuliu/MS-G3D is open source, released under the MIT license.
- What language is MS-G3D written in?
- kenziyuliu/MS-G3D is primarily written in Python.
- How popular is MS-G3D?
- kenziyuliu/MS-G3D has 459 stars on GitHub.
- Where can I find MS-G3D?
- kenziyuliu/MS-G3D is on GitHub at https://github.com/kenziyuliu/MS-G3D.