Research on face tracking system based on YOLOv5s

Some surveillance equipment can detect faces, but when the face is not in the picture, or when the face is incomplete there may be a comparison error, in order to reduce this comparison error requires the camera to track the face in real time, and subject to cost, technology and other issues, nowadays face tracking technology is less used in these devices. Therefore, this paper proposes a low-cost face recognition and tracking system, so that the YOLOv5 algorithm and trained face data set to, combined with the STM32 core board and servo to realize, face recognition and tracking. The realization results show that. The accuracy of face recognition reaches 95% and can accurately recognize faces in the corresponding database.