A Real-Time Suspicious Stay Detection System Based on Face Detection and Tracking in Monitor Videos

摘要

This paper proposed a novel approach of a suspicious stay detection system based on face detection and tracking. Through the intelligent analysis of real-time monitor videos, we judge whether there is any suspicious stay in the monitoring area. In the proposed system, we firstly set the face detection and tracking area. Then we use the convolution neural network algorithm to detect the face and the Particle Filter to track the face. Finally, the time someone stays in the surveillance area is used to determine whether he is stuck suspiciously. Experimental results show that the proposed approach has a relatively high accuracy, and meets the real-time requirement.