Self-trained prediction model and novel anomaly score mechanism for video anomaly detection

作者：

Highlights：

• The average prediction losses of the pseudo normal and anomalous frames are used to calculate anomaly score.

• The application of self-training mechanism eliminates the need for manually labeled training data.

• The proposed unsupervised method adopts the idea of anchors and achieves effective anomalous region localization.

摘要

•The average prediction losses of the pseudo normal and anomalous frames are used to calculate anomaly score.•The application of self-training mechanism eliminates the need for manually labeled training data.•The proposed unsupervised method adopts the idea of anchors and achieves effective anomalous region localization.

论文关键词：Anomaly detection,Unsupervised method,Memory module,Reconstruction,Self-training mechanism

论文评审过程：Received 28 July 2021, Revised 6 January 2022, Accepted 16 January 2022, Available online 31 January 2022, Version of Record 14 February 2022.

论文官网地址：https://doi.org/10.1016/j.imavis.2022.104391