Human action recognition in RGB-D videos using motion sequence information and deep learning

作者：

Highlights：

• An approach to recognize human actions in RGB-D videos using motion sequence information and deep learning is proposed.

• Proposed a new representation of motion information for human action recognition that emphasizes motion in various temporal regions.

• The use of motion information in RGB and depth video streams.

• Analysis using t-SNE visualization of ConvNet features to show the discriminative characteristics of the proposed representation.

摘要

•An approach to recognize human actions in RGB-D videos using motion sequence information and deep learning is proposed.•Proposed a new representation of motion information for human action recognition that emphasizes motion in various temporal regions.•The use of motion information in RGB and depth video streams.•Analysis using t-SNE visualization of ConvNet features to show the discriminative characteristics of the proposed representation.

论文关键词：Multi-modal action recognition,Deep learning,Motion information,Extreme learning machines

论文评审过程：Received 15 December 2016, Revised 5 July 2017, Accepted 7 July 2017, Available online 8 July 2017, Version of Record 17 August 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.07.013