COVID 19: Identification of Masked Face using CNN Architecture
Tackling with the Covid-19 pandemic has been a grueling task from the day one of its rise. To prevent the spread of the virus, one of the major steps taken by governments of the world is to mandate the use of face masks in public places. This has made face recognition process in today’s era quite a hideous process as masked faces are difficult to analyze because most of the key parts of the face get covered by the mask. Thus, it raises a matter of great concern when security and individual identification comes into play. Therefore, this paper proposes a reliable system to recognize the individual even when the face is covered with mask. The first step aims to convert all the images of the database to grayscale and then convert them into vectors. This reduces the contributing factor of the masked region in the prediction process. The vectors are then fed to a Convolutional Neural Network. A sequential model is defined to design multiple layers of the network and generate output. The proposed model showed maximum accuracy of 90% after tuning the key parameters of the model.