SGM-Net: Skeleton-guided multimodal network for action recognition

作者：

Highlights：

• We propose skeleton-guided multimodal network (SGM-Net).

• The guided block in SGM-Net is proposed to obtain action-related object feature.

• Two schemes of correlation operation in guided block are investigated.

• The experimental results demonstrate that our method achieves state-of-the-art performance on NTU and Sub-JHMDB datasets.

摘要

•We propose skeleton-guided multimodal network (SGM-Net).•The guided block in SGM-Net is proposed to obtain action-related object feature.•Two schemes of correlation operation in guided block are investigated.•The experimental results demonstrate that our method achieves state-of-the-art performance on NTU and Sub-JHMDB datasets.

论文关键词：Action recognition,multi-modality,skeleton-guided

论文评审过程：Received 31 May 2019, Revised 24 March 2020, Accepted 29 March 2020, Available online 9 April 2020, Version of Record 27 April 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107356