Correlation Net: Spatiotemporal multimodal deep learning for action recognition

作者：

Highlights：

• The proposed model captures spatiotemporal correlation without time correspondence.

• Introduce Shannon fusion to select features based on distribution entropy.

• The proposed network provides complementary information for long video recognition.

摘要

•The proposed model captures spatiotemporal correlation without time correspondence.•Introduce Shannon fusion to select features based on distribution entropy.•The proposed network provides complementary information for long video recognition.

论文关键词：Correlation Net,CNN,Activity recognition,Deep learning,Fusion

论文评审过程：Received 4 May 2019, Revised 2 December 2019, Accepted 2 December 2019, Available online 10 December 2019, Version of Record 14 December 2019.

论文官网地址：https://doi.org/10.1016/j.image.2019.115731