Real-time Abnormal Behavior Recognition and Monitoring System Based on Panoramic Video

Abstract

Traditional monitoring systems are mostly monocular cameras, the main function of which is to play and store video in the monitoring area and to achieve the collection and judgment of the information in the video through artificial observation. Usually, emergencies that are not found and processed in time will cause serious situations. Compared with the monocular camera, the panoramic camera has a wide field of view and no blind spots. This system uses a panoramic camera to achieve 360-degree video acquisition, and uses deep learning to process the video to obtain the posture of the human body, and then processes it through the human behavior recognition network to identify the abnormal human behavior in real time. Through the dual warnings of local and remote clients and the artificial re-judgment of the received abnormal behavior video, the manager can timely find and handle abnormal behavior. This system not only saves labor costs, but also improves the utilization efficiency of video. The 360-degree all-round perspective provides the area manager with an immersive environment reproduction, allowing him to grasp the details of the scene in real time. This system provides virtual on-site conditions for decision-making and handling of emergencies, enabling early warning and minimizing losses.