Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition

[joyhuang9473]

Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition

• Author: Sijie Yan, Yuanjun Xiong, Dahua Lin
• Origin: https://arxiv.org/abs/1801.07455
• Related:
  • https://github.com/yysijie/st-gcn/blob/master/OLD_README.md
  • https://github.com/open-mmlab/mmskeleton/blob/master/doc/START_RECOGNITION.md

we propose a novel model of dynamic skeletons called Spatial-Temporal Graph Convolutional Networks (ST-GCN), which moves beyond the limitations of previous methods by automatically learning both the spatial and temporal patterns from data.