Abnormal Activity Detection Techniques in Intelligent Video Surveillance: A Survey

Abstract

Currently, CCTV (Closed Circuit Television) cameras are used for surveillance by alerting the security officer if any malfunction or abnormal activity happens. Abnormal activities may be theft, violence, or explosion. CCTV cameras are used in public places like city streets, parks, communities, and neighborhoods to help detect crime and enhance public safety. Manual surveillance for this is tedious and time-consuming. Detecting abnormal crowd behavior in real-time is an exciting research area. Presently, most researchers are interested in developing Dynamic abnormal detection mechanisms to ensure security. However, this is challenging due to climate change, human movement, occlusions, and low video quality. Due to the high dimensionality of video data, Space and time complexity are also increased. This paper explains the various methods of abnormal activity detection under deep learning and the handcrafted approach.