Fast sparse coding networks for anomaly detection in videos

作者：

Highlights：

• Novel methods focus on high-level features rather than frame reconstruction error.

• Discriminative Spatial-Temporal Fusion Features for anomaly detection in videos.

• Fast Sparse Coding Networks achieve higher accuracy at maximum 10000 lower latency.

• Experiments show the superiority of our method in accuracy and efficiency.

摘要

•Novel methods focus on high-level features rather than frame reconstruction error.•Discriminative Spatial-Temporal Fusion Features for anomaly detection in videos.•Fast Sparse Coding Networks achieve higher accuracy at maximum 10000 lower latency.•Experiments show the superiority of our method in accuracy and efficiency.

论文关键词：Anomaly detection,Encoding-decoding networks,Sparse coding networks,Spatial-temporal information,Video representation

论文评审过程：Received 29 August 2019, Revised 6 March 2020, Accepted 22 June 2020, Available online 24 June 2020, Version of Record 29 June 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107515