Enhancing COVID-19 Safety: Automatic Mask Detection and Temperature Monitoring System using YOLOv5x and Mobile App Platform

COVID-19 has several effects and disruptions to all sides of daily life. To minimize this effect, face mask should be worn to control its dangerous. To avoid the manual personal checking of the mask, which is a difficult and disturbing issue, the suggested idea is to develop an auto mask-checking based on mobile phone application platform that could be continuously running in real-time leading to a more beneficial and cheap surveillance system. Quite modern object detection platform, called, you only look once version 5x (yOLOv5x) is considered to realize this function. The model is trained using a custom -created dataset grouping mask/unmask persons' pictures. The model is invoked in Google Colab, transformed to a mobile application based on a light version of YOLOv5x, and then tested in real-time on a mobile phone. The application has also the capability for checking facemasks for a group of individuals in real-time with high-speed inference. Body temperature is also incorporated for each person using IR touchless sensor and monitored on the application interface. The developed real-time detector system is small-sized, low cost, lightweight, and fast response time, reaching 98.8% training\testing accuracy rate.