Detection and Recognition of Abnormal Behavior based on Multi-level Residual Network

The ability of real-time detection and recognition of abnormal behavior in video monitoring system is a key problem in intelligent monitoring system. This paper proposes a network framework based on multi-level residual network to detect and recognize abnormal human behavior from video. The framework of multi-level residual network includes human body detection module and posture recognition module. Based on the former, this paper proposes the detection residual network (d-Res) to adopt multi-scale target detection strategy to ensure the detection speed and detection effect of human body. The latter is used to extract spatial features of abnormal behaviors, and the recognition residual network (r-Res) based on transfer learning is used to extract deep features of images, so as to classify abnormal behaviors efficiently. Experiments are carried out on UTI dataset to evaluate the performance of the proposed algorithm. The results show that the proposed method is effective in detecting and recognizing abnormal behaviors in real-world scenes.