Face Mask Detection and Social Distancing using Deep Learning

Abstract

Social distancing and wearing a face mask correctly is known to be one of the most effective measures to fight against a pandemic like Covid 19. Thereupon no such precise system has been made and in this domain, research is still going on. In this study, mainly two deep learning models namely CNN, and YoloV5 are employed for object detection of face masks and social distancing and Vgg-19 for feature extraction. For the evaluation of the models, various parameters like precision, recall, mAP-mean average precision, accuracy, validation and training loss have been calculated. This has been observed that among all deployed deep learning models on the collected data, CNN (Convolutional Neural Network) outperformed with an accuracy of 99.3% and a precision of 98%.