Research on Target Detection of Regional Monitoring with Complex Background using CNN and Background Modelling

Abstract

The regional monitoring systems aim to recognize and localize the target of interest in the region area. However, the target detection algorithm currently used in the regional monitoring system has problems such as low recognition probability under complex background conditions. The type of moving object recognized by the background modelling algorithm is difficult to judge. This paper proposes a monitoring area target detection method that fuses the detection results of the two YOLOv5 target detection algorithms and the Vibe background modelling method through Kalman filtering. Experiments show that the proposed method can improve the consistency and stability of target detection results in regional monitoring scenarios.