Automatic Face Mask Detection Using a Hide and Seek Algorithm

Pratyaksh Bhalla, Soumya Snigdha Kundu, S. Deepanjali, G. Vadivu, Sapdo Utomo

Abstract: The coronavirus has affected millions around the world and has inevitably brought about a necessity to wear face masks in official and public places to take the first step in keeping one’s self safe. To monitor personnel and public areas and prevent the spread of the disease we present a scalable and deployable face mask detection system in a real time setting using a novel hide and seek algorithm. Our model, based on openCV library and dlib environment utilizes the facial landmarks where in the algorithm detects face masks through the presence and absence of facial markers. We call this process as seeking and hiding. We overcome present issues of high computational cost of deep learning models and low inference speeds of general detection paradigms. We also validate our algorithm on several aspects which affect the accuracy of other models such as image and face orientation, type of face masks and more. As our model requires no data for model training, we eliminate the highly sensitive issue of acquiring facial data and bias. Our model achieves 98.79\(\%\) precision and 94.81\(\%\) recall.