Abstract

A methodical approach to curb the spread of this menacing disease, COVID-19 must be taken. 'CO' answer for corona, 'VI' denote virus, and 'D' represent disease. The absence of highly effective medicines and scarcity of vaccination make this disease more lethal and vicious; this makes it important for us to find a provisional yet efficacious way to cushion ourselves and that one love. Wearing masks can act like that cushion, it's truly a camouflage, acting as a Non-Pharmaceutical Intervention (NPI) proceeding that could be easily implemented without much capital investment. This thesis evaluates an efficient way of face mask detection that can be used by private or government authorities as a tool against COVID-19. This research aims to build a light weighted and user-friendly model that can be easily used in static or real-time face mask detection. This study of face mask detection is made possible using Deep Learning, Convolutional neural network algorithm, and MobileNetV2, a python open-source image-processing and classification model. Steps involved in designing and implementing the model are collecting and accessing the dataset, data processing and encoding, testing, and training data, accuracy prediction, and model implementation in a real-time project that can instantly detect and provide the desired output. The model can make 98% accurate production in real-time, distinguishing between individuals with a mask or without ma masks. Our proposed face mask detection model outperforms existing models in terms of accuracy and easy usability.