Face mask detection on single-face and multi-face video using convolutional neural network

Nowadays, coronavirus disease has become more widespread. The virus is able to attack the human respiratory function. One of the ways to decrease the virus spread is to wear a face mask correctly in a public place. This paper proposed a RetinaFace for detecting a face and Convolutional Neural Network for classifying a face mask. The experimental dataset contains single-face and multi-face videos recorded using a drone camera. Based on the test results, NASNetMobile architecture yields the best performance with accuracy, precision, recall, and f1-score of 82.76%, 100%, 78.72%, 88.09%, respectively.