Face Mask Detection using Single Shot Multibox Detector

Abstract

In 2019, COVID has originated in Wuhan, China and has rapidly proliferated across the globe. Since then, various governments have taken up measures such as Mask mandates, Social Distancing, frequent sanitization and providing emphasis on hygiene to contain the virus. In order to aid this effort and to identify citizens with facetious attitude towards emergency measures, who frequently refuse to wear face masks, we can use deep learning to help mitigate  this issue. Object detection has robust libraries of algorithms such as YOLO, SSD and R-CNN to detect face occlusions via consecutive feature  extraction  through  convolution  layers and feeding it to a detector. The experimental results  have been  done  in  real-time  application  and  the  device  has   been installed domestically for indoor usage.  The  experimental results shows that this technique can identify individuals who wear or do  not  wear  the  face  mask  precisely regardless of whether they are moving to a different position.