Système de Détection du Port de Masque dans les Salles de Cours par Deep learning (Application Implémentée sur Raspberry pi)

Abstract

Since 2019 the world has been facing a huge health crisis due to the rapid transmission of the coronavirus (COVID-19). Several guidelines have been issued by the World Health Organization (WHO) for protection against the spread of the coronavirus. According to the WHO, the most effective preventive measure against the virus is to wear a mask in public places and crowded areas. However, it is very difficult to monitor people in these areas. The researchers have proposed an intelligent solution to implement smart cameras in places (public and closed) capable of detecting and reporting people who are not wearing masks. Our project involves designing an intelligent system for detecting mask-wearing in class rooms based on deep convolutional networks (DCNN). The implemented system is embed ded in Raspberry Pi 3. It was built by refining the “Inception V3” deep learning model trained on 12K Images dataset containing more than 12,000 images of three classes of people : ( 1) with masks worn correctly, (2) with masks worn incorrectly, and (3) without masks. The tests were carried out on images taken in classrooms (Amphitheatres), tutorial rooms, and practical work rooms. The results obtained are encouraging, but a refinement of the learning, especially in the class on badly worn masks, will make it possible to reduce the errors of detection and confusion in this class and consequently boost the reliability of our system.