CNN-Based Model for Real-Time Identification of Face Masks

A global pandemic known as COVID-19 emerged in 19 countries across the world. Wearing a facial covering can be effective in preventing the airborne transmission of contagious pathogens and curbing the spread of infections. An automated system for recognizing whether an individual is wearing a mask has been introduced in this research. This study introduces an innovative approach to facial mask detection using convolutional neural networks (CNN). Initially, the Haar-Cascade technique is employed for image identification. When combined with contemporary algorithms, this classifier demonstrates a remarkable ability to identify faces across various expressions, making effective use of feature selection and minimizing false positive identifications. The Haar feature-based cascade classifier method uses just 200 out of 6000 attributes, achieving a recognition rate of 85–95%. The need for a mask detection system is crucial during the global COVID-19 pandemic as a specialized public health service. Training the model relies on a dataset consisting of images of both masked and unmasked faces. The performance of proposed approach is assessed using the different parameters such as accuracy, precision, recall, specificity, F1-score, and kappa coefficient. From the result it can be said that proposed CNN model is providing better accuracy, reduced losses and predicts user with different labels such as incorrect mask, without mask, and with mask.