An Implementation of Face Mask Detection System Based on YOLOv4 Architecture
Zhaoxiang Qin, Zheng Guo, Yuzhe Lin · 2022
To solve the problems of low intelligence, low real-time, and poor warning in the face of a complex environment, a face mask detection system based on YOLOv4 architecture is implemented by combining the face mask dataset in the real world. The system can correctly detect faces from images and real-time videos, and judge whether the target face is wearing a mask in a standard way. From the experimental results, the system can distinguish the detected faces with three labels as with a mask, without a mask, and mask worn incorrectly, and achieve a mean Average Precision (mAP) of 87.94%. At the same time, the system can realize the dual combination of detection and warning by recording and uploading face data, which is of high academic research value and practical application value.