Multi-stream slowFast graph convolutional networks for skeleton-based action recognition

作者：

Highlights：

• A SlowFast graph convolution network is proposed for skeleton action recognition.

• The main advantage of this work is low computational cost and strong performance.

• Six streams of skeleton sequences are fused for better recognition accuracy.

• Two attention modules enhance the spatiotemporal modeling ability.

摘要

•A SlowFast graph convolution network is proposed for skeleton action recognition.•The main advantage of this work is low computational cost and strong performance.•Six streams of skeleton sequences are fused for better recognition accuracy.•Two attention modules enhance the spatiotemporal modeling ability.

论文关键词：Action recognition,Graph convolutional network,Human skeleton,SlowFast network,Attention

论文评审过程：Received 20 June 2020, Revised 19 January 2021, Accepted 16 February 2021, Available online 23 February 2021, Version of Record 4 March 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104141