Detecting Face Mask for Prevent COVID-19 Using Deep Learning: A Novel Approach

ABSTRACT

The recent times we have seen that not taking proper measures like wearing mask or sanitization is the leading cause of the spread of “COVID-19”. Despite so many rules and regulations around the globe regarding proper sanitization and wearing of a mask, many people tend to ignore wearing of facemask when they are in public places. Although vaccines like covishield, covaxin, AstraZeneca, Pfizer-BioNTech, sputnik, etc. have been developed and the no of cases are decreasing in some countries to stop further transmit of virus it is still necessary to wear a mask in public places. Technology to monitor whether a person is wearing a mask in a public place or not is needed. One potential device like a CCTV camera can be used in this case. We came up with an algorithm “MaskYolo” which can be integrated with CCTV cameras that can detect if a person is wearing a mask or not using You Only Look Once (YOLO) algorithms. For our work, we used YOLOv4 and compared it to its sibling YOLOv4 Tiny. The best precision reached was 93.9 for YOLOv4 and 88.7 for YOLOv4 Tiny. Overall YOLOv4 stands out in all aspects for our model “MaskYolo”. Thus, we can use “MaskYolo” and build a device that detects if a person is wearing a mask or not. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.