Method for detecting illegal call answering and making behaviors in restricted area

Abstract

The embodiment of the invention discloses a behavior detection method for illegal call receiving and making in a restricted area, and the method comprises the steps: collecting a video image of a monitoring area, carrying out the foreground and background separation of the video image, carrying out the human body recognition of an obtained foreground image through a trained network model, and extracting the multi-domain features of a human body, and human body behaviors are identified, and early warning of a corresponding level is made. The trained network model is obtained by training a YOLO network model by using a comprehensive multi-angle sample library expanded by a multi-view generative adversarial network model. According to the method, the human body in the restricted area is identified by expanding the multi-view different-distance sample library, the multi-domain features of the human body are extracted, and the multi-domain features of the human body are used for reasoning to realize human body behavior identification and early warning level detection, so that the purpose of detecting illegal call making and receiving in the restricted area is achieved.

Description

technical field [0001] The present application relates to the technical field of video monitoring, in particular to a method for detecting illegal phone calls in a restricted area. Background technique [0002] Video surveillance plays a key role in maintaining public safety and social stability. With the rapid development of military, commercial and other industries, security video surveillance is changing towards more complex application scenarios and more accurate monitoring results and requirements and standards. The existing image recognition of phone calls is mainly aimed at the relevant personnel driving vehicles, and the distance and angle between the human body and the camera are relatively stable. However, the human body in the restricted area is moving, and changes in the relative angle and distance between the human body and the camera directly affect the human behavior monitored in the collected video, which in turn affects the accuracy of abnormal behavior rec...

AI Technical Summary

Problems solved by technology

However, the human body in the restricted area is moving, and changes in the relative angle and distance between the human body and the camera directly affect the human behavior monitored in the collected video, which in turn affects the accuracy of abnormal behavior recognition. Behaviors such as making phone calls are covered to a certain extent

This brings two major difficulties to the identification of illegal phone calls in restricted areas: 1) The camera captures the human body from various angles and the proportion of the frame is different, which brings a huge challenge to the accuracy of target recognition. 2) The camera is facing away from the camera. The existence of occlusion when making and receiving calls puts new requirements on the recognition of behaviors

Images