RESEARCH ON DEEP LEARNING NETWORK YOLOv5 TO RECOGNIZE NOT WEARING FACE MASK IN A VIDEO

Abstract

On 31 December 2019, WHO was informed of cases of pneumonia of unknown cause Wuhan City, China. A novel coronavirus was identified as the cause by Chinese authorities on 7 January 2020 which is now known as SARS-CoV-2. Since it first appeared, COVID-19 has been spreading rapidly, inflicting significant harm and infecting hundreds of millions of people globally. Although there is no particular remedy, many nations use some temporary policies as a solution to limit the transmission of the disease such as physical or social distancing, quarantining, ventilation of indoor spaces, covering coughs and sneezes, hand washing and keeping unwashed hands away from the face. Studies show that COVID-19 is mostly passed from person to person through respiratory droplets when a person cough, sneeze, talk, shout, or sing. These droplets can subsequently land in the mouths or noses of people who are nearby or they may breathe these droplets in. Many researches claim that wearing a mask over the nose and mouth reduces the discharge of droplets, hence, it is a simple barrier to help prevent respiratory droplets from reaching people. Therefore, the use of face masks or covers has been obligated in most of public places which has resulted in an increase in demand for automatic real-time mask detection systems to replace manual reminding. In addition, deep learning and computer vision are being strongly researched. Especially the development of YOLO family of models in recent years bodes well for many current artificial intelligence projects. The most recent version of YOLO base network is YOLOv5, which was developed by Ultralytics LLC. According to the data supplied by the company this version is now extremely promising and seem to be the most powerful object detection algorithm at present. From those importance and necessity of wearing mask in the anti-epidemic time as well as potentials of YOLOv5, this project “Research on Deep Learning Network YOLOv5 to Recognize Not Wearing Face Mask in a Video” is run in order to detecting and classifying whether faces in public areas are wearing masks correctly. The experimental results show that the algorithm proposed in this paper can effectively identify face mask and might be apply in the actual environment for supervision purpose.