Non-perception face image acquisition method and system

Abstract

The invention discloses a non-perception face image acquisition method and system, and the method comprises the steps: S1, photographing a panoramic image of a to-be-detected region, detecting the position of a human body in the panoramic image through a human body detection method, and obtaining a human body bounding box vector bi; s2, transmitting the human body bounding box vector bi to a pre-trained PTZ camera motion model, wherein the PTZ camera motion model is used for outputting a PTZ camera motion parameter vector which corresponds to the input human body bounding box vector bi and canshoot a human front face image; and S3, the PTZ camera moves the direction of the camera and changes the focal length of the camera according to the PTZ camera motion parameter vector, aligns at a person at a corresponding position for shooting, realizes face angle screening through face detection and head posture recognition algorithms, and acquires a front face image of the person. A two-stagedetection method combining human body detection and face angle screening is adopted, the face detection efficiency is high, and the practicability is high.

Description

technical field [0001] The present invention relates to the field of deep learning and computer vision, in particular to a non-perceptual face image acquisition method and system. Background technique [0002] In some scenarios, such as students attending classes in a classroom, and members of an organization meeting in a room, it is necessary to verify the identity of the people in the room. When all personnel are facing the same direction, such as facing a wall of the room (blackboard, projection screen), deploying a camera at a suitable position can collect facial images of all personnel in the room, and applying face recognition related algorithms can realize non-perceptual classroom Personnel attendance. [0003] In the actual classroom environment and conference environment, people are densely populated and people’s postures are changeable. The existing face image collection methods have a high rate of missed detection, and the quality of the extracted face images is ...

AI Technical Summary

Problems solved by technology

However, relatively few studies have applied dual cameras to person recognition

Related research has used the method of dual-camera classroom inspection to identify attendance students. The comparative document CN110647842A discloses an invention patent named "A Dual-camera Classroom Inspection Method and System". The coordinates of the center point control the second camera to monitor the students in front of each desk at a fixed point. The disadvantage of this patent is that it does not make full use of the information captured by the panoramic camera, and it is difficult to obtain the faces of all students. There are cases of missed detection, and Additional cruise routes need to be designed, which is inefficient and wastes detection resources and time

Images