Artificial intelligence application to monitor the use of protective masks

Abstract

This work presents a web application that monitors face mask use in public environments in the current pandemic context. The application was executed using the Flask framework and the Python language. It has a control panel that helps the user visualize the data obtained. The monitoring process uses the Haar Cascade algorithm to classify faces with and without face masks. The web application is lightweight. It allows the detection and storage of the captured images on the “cloud” and the possibility of further data analysis. The classifier has a precision, revocation, and f-score of 63%, 93%, and 75%, respectively. Although the precision was satisfactory, further experiments will be conducted to explore new computer vision techniques, such as deep learning.