Deep Learning Based People Detection, Tracking and Re-identification in Intelligent Video Surveillance System

Intelligent video surveillance has gained increasing needs and attentions due to its greatly potential in security and social governance. In this paper, we integrate three parts of video surveillance system together, including people detection, tracking and person re-identification to construct a systematic research framework. Following a brief introduction, we first present current problems and difficulties of people detection, tracking and re-identification. Then the state-of-the-art deep learning based algorithms in recent years are summarized and evaluated. At last we put forward the prospect of current research directions, providing a relatively comprehensive knowledge base for the future development of intelligent surveillance video systems.