Application of People Flow and Face Mask Detection for Smart Anti-Epidemic.

During the COVID-19 pandemic, World Health Organization (WHO) published two protection methods. One is maintaining social distance; the other is wearing the mask. Many reports indicate that keeping a safe distance and wearing face masks can reduce the risk of transmission during this pandemic. In this paper, we present a deep learning pipeline that can achieve pedestrian flow statistics, identify correct and incorrect mask-wearing simultaneously from real-time video streams. In our proposed system, the live video stream is taken as input. The alert will be broadcasted when people crowed and someone do not wear the mask. Experimental results show that the proposed system can achieve an effective detection solution and reduce the risk of infection.