Enhancing Public Health Measures with AI and MobilenetV2 for Face Mask Detection

All continents were experiencing a global health crisis brought on by the coronavirus (COVID-19). To properly protect against this infection, it is important to put preventive measures into place. The wearing of face masks in public places is a reliable form of protection. A big issue arises when attempting to detect the presence of masks among huge groups in public areas. The existing face mask detection systems are concisely described in this work. Based on the methodology used, the databases used, and the accuracy attained, we compared the state of the art of face mask detection techniques. This study brought to light the difficulties that still remain in creating a reliable face mask detection system. A two-step system was formed as a solution to this problem. A model was trained in the first stage using MobileNetV2, OpenCV, and Keras/TensorFlow. The training set included both masked and unmasked faces from real-world data as well as data from GitHub and Kaggle. This model achieved a 99.35% accuracy rate. Deep learning methods, especially when optimized, are effective in a wide range of scenarios, including complex and dynamic environments. Hybrid methods can be effective when a combination of techniques is beneficial, such as handling diverse mask types