A Comprehensive Evaluation of Face Mask Identification System using Modified Deep Learning Methodology

Recent years have seen an explosion in the prevalence, infectiousness, and potential harm posed to humanity by the corona virus family of viruses. When one person breathes out the virus, it lands on surfaces and other people breathe them in, causing the sickness to spread from person to person. As a result, safeguarding ourselves and those close to us has assumed paramount importance. There are measures we may do to protect ourselves, including as keeping our distance from others, wearing a mask, washing our hands often (every two hours), and using hand sanitizer. Masks worn by the general public have recently grown ubiquitous across the globe. The situation in India is particularly severe because of the country's dense population packed into a relatively limited landmass. Using a cross-validation strategy involving the conventional learning model known as Convolutional Neural Network (CNN), this paper presents a method called Modified Learning based Mask Identification Rule (MLMIR) for detecting whether or not a face mask is worn in offices or other places with a high volume of employees. The model's performance using live video streaming was astonishing once it was trained with a dataset made from real-world data. Using a number of additional parameters and a large number of people placed at different heights and angles inside the picture, we further evaluate the model's accuracy.