Effective Facemask Detection Using a Few Learning-Based Recognition Methods

Global healthcare systems are effectively dealing with the pandemic arising out of COVID-19 virus yet there are requirements to formulate appropriate risk minimization methods. In the absence of effective medical resources, certain alternatives are recommended to stem the infection. Mask wearing is regarded to be a non-pharmaceutical intervention measure to prevent rapid spread of the virus from an infected individual. As part of measures to ensure wearing of marks in public places, certain methods of monitoring are being developed. Majority of these are based on learning-aided pattern recognition methods. This paper discusses the design of a few learning-based methods like Artificial Neural Network (ANN), Convolutional Neural Network (CNN) and Support Vector Machine (SVM) and also includes discussion on the performance of these methods which are configured and trained to find the best suitable approach for checking mask wearing in a real-time situation. It is observed that the proposed technique based on VGG-16 CNN achieves high accuracy (99.0%) during the testing phase but when implemented with SVM during training, similar results are obtained.