Identity recognition method for illegal object based on illegal behavior recognition

Abstract

The invention provides a violation object identity recognition method based on violation behavior recognition, and the method comprises the steps: obtaining and caching a monitoring video in a preset scene in real time, judging whether a violation behavior exists in the preset scene, recognizing the identities of all target objects in the preset scene when the violation behavior is detected, and carrying out illegal behavior secondary judgment on all target objects in the preset scene. The method is used for monitoring whether the illegal behavior exists in the preset scene or not, identifying the identity of the corresponding personnel when the illegal behavior is detected to exist, carrying out secondary judgment on the illegal behavior on the determined personnel identity, and controlling whether to carry out alarm prompting or not according to the judgment result, so that the functions of accurately monitoring whether the illegal behavior exists in the preset scene or not and determining the identity of the corresponding personnel are realized.

Description

technical field [0001] The present invention relates to the technical field of identification methods, in particular to an identification method of an illegal object based on the identification of illegal behaviors. Background technique [0002] At present, taking the safety helmet detection on the construction site as an example, considering the computing power and performance, the current visual detection scheme is to detect the frame of the surveillance video screen, that is, to extract pictures from the video screen at a certain frequency, and then detect the Whether the personnel wearing helmets, but because it is a frame, it may mean that when a picture without a helmet is recognized, it may not meet the conditions for face recognition (the face is not clear, the face tilt angle is too large) Large, face occlusion, etc.), so further face recognition cannot be performed, and the identity of the person cannot be confirmed. In the existing technical solutions, because th...

AI Technical Summary

Problems solved by technology

[0002] At present, taking the safety helmet detection on the construction site as an example, considering the computing power and performance, the current visual detection scheme is to detect the frame of the surveillance video screen, that is, to extract pictures from the video screen at a certain frequency, and then detect the Whether the personnel wearing helmets, but because it is a frame, it may mean that when a picture without a helmet is recognized, it may not meet the conditions for face recognition (the face is not clear, the face tilt angle is too large) large, face occlusion, etc.), so further face recognition cannot be performed, and the identity of the person cannot be confirmed

The existing technical solutions, because the frame detection is performed at a certain frequency, may not be able to provide a picture of the violator that can identify the identity of the face, and it is impossible to further realize the face recognition of the identity of the violator

Images