Mining human movement evolution for complex action recognition

作者：

Highlights：

• A mid-level feature is proposed to obtain temporal structural information in videos.

• A hierarchical video representation is developed to encode videos at multiple levels.

• A newly developed computational framework of human action recognition is presented.

摘要

•A mid-level feature is proposed to obtain temporal structural information in videos.•A hierarchical video representation is developed to encode videos at multiple levels.•A newly developed computational framework of human action recognition is presented.

论文关键词：Action recognition,Dense trajectory,Motion compensation,Feature representation,Hierarchical encoding

论文评审过程：Received 9 November 2016, Revised 5 January 2017, Accepted 9 February 2017, Available online 10 February 2017, Version of Record 21 February 2017.

论文官网地址：https://doi.org/10.1016/j.eswa.2017.02.020