Recognizing the Type of Mask or Respirator Worn Through a CNN Trained with a Novel Database

Abstract

Since the onset of the coronavirus pandemic, researchers from all over the world have been working on projects aimed at countering its advance. The authors of this paper want to go in this direction through the study of a system capable of recognizing the type of mask or respirator worn by a person. It can be used to implement automatic entry controls in high protection areas, where people can feel comfortable and safe. It can also be used to make sure that people who work daily in contact with particles, chemicals, or other impurities wear appropriate respiratory protection. In this paper, a proof-of-concept of this system will be presented. It has been realized by using a state-of-the-art Convolutional Neural Network (CNN), EfficientNet, which was trained on a novel database, called the Facial Masks and Respirators Database (FMR-DB). Unlike other databases released so far, it has an accurate classification of the most important types of facial masks and respirators and their degree of protection. It is also at the complete disposal of the scientific community.