基于深度学习的异常事件检测研究综述

Survey on Deep Learning Approaches for Anomalous Event Detection

基于视频的异常事件自动检测技术一直是学术界和企业界研究的热点,在人群监控、交通管理、老人儿童和特殊病人的护理等方面发挥了重要的作用。文章首先介绍传统的异常检测方法,然后从评价指标、基本网络模型、检测框架、基准数据集等方面详细介绍了深度学习在异常检测中的研究进展,最后对未来的研究方向进行了展望。

The technology of of video-based abnormal events automatic detection has always been a hot topic in academic and business circles,which plays an important role in crowd detection,traffic management,nursing of the elderly and children and special patients.In this paper,firstly we introduces the traditional anomalous detection methods,then the research progress of deep learning in anomaly detection is introduced in detail from evaluation index,basic network model,detection framework,benchmark data set and so on.Finally,the future research direction is prospected.