Intelligent video anomaly detection and classification using faster RCNN with deep reinforcement learning model

作者：

Highlights：

• Design an intelligent IVADC-FDRL model for video anomaly detection and classification.

• Propose a Faster RCNN with ResNet as a baseline model for Anomaly Detection.

• Employ a deep Q-learning (DQL) model to classify detected anomalies in video frames.

• Validate the anomaly detection and classification performance on UCSD Anomaly dataset.

• Proposed model achieves a maximum accuracy of 98.50% on the applied Test004 dataset.

摘要

•Design an intelligent IVADC-FDRL model for video anomaly detection and classification.•Propose a Faster RCNN with ResNet as a baseline model for Anomaly Detection.•Employ a deep Q-learning (DQL) model to classify detected anomalies in video frames.•Validate the anomaly detection and classification performance on UCSD Anomaly dataset.•Proposed model achieves a maximum accuracy of 98.50% on the applied Test004 dataset.

论文关键词：Video surveillance,Intelligent systems,Anomaly detection,Deep reinforcement learning,UCSD dataset

论文评审过程：Received 1 April 2021, Revised 18 May 2021, Accepted 1 June 2021, Available online 2 June 2021, Version of Record 12 June 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104229