A study on attention-based LSTM for abnormal behavior recognition with variable pooling

作者：

Highlights：

• Realizes the transformation from 2D skeleton to 3D skeleton through geometric methods.

• Combines spatial and temporal information from multiple skeletons.

• Dynamically handle the behavior recognition of multi-person skeleton sequences.

• The proposed framework is evaluated with real-world surveillance video data.

摘要

•Realizes the transformation from 2D skeleton to 3D skeleton through geometric methods.•Combines spatial and temporal information from multiple skeletons.•Dynamically handle the behavior recognition of multi-person skeleton sequences.•The proposed framework is evaluated with real-world surveillance video data.

论文关键词：Abnormal behavior,Attention,LSTM,Variable pooling

论文评审过程：Received 31 May 2020, Revised 6 December 2020, Accepted 27 January 2021, Available online 4 February 2021, Version of Record 16 February 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104120