Enhancing Public Health: a Better Approach for Face Mask Detection Using Transfer Learning to Prevent Airborne Disease

Abstract

Face mask typically refers to a covering that is worn over the nose and mouth to provide protection from airborne particles and potentially harmful substances. The primary purpose of a face mask is to reduce the transmission of respiratory droplets that may contain viruses, bacteria, or other contaminants, especially in situations where maintaining physical distance is important. Computer Vision can help to monitor the use of face masks based on images captured via CCTV. A previous study built a mask detection system using Convolutional Neural Networks (CNN) based models, which produced high accuracy but was limited to the front face. This research focuses on leveraging computer vision and machine learning, Deep learning techniques for accurate face mask detection. The proposed approach employs transfer learning, utilizing MobileNetV2 as the base model, coupled with a custom classifier. This model consists of two core components: face detection (faceNet) and face mask classification (maskNet), following established machine learning and deep learning workflows. Experimental results underscore its effectiveness, achieving a remarkable 97.87% accuracy in identifying individuals wearing masks, 98.46% accuracy in detecting those without masks, culminating in an impressive overall model accuracy of 98.33%. In addition to its primary role in monitoring mask compliance, this research highlights its potential to make meaningful contributions to technological progress and endeavors aimed at enhancing public health.