Face Mask Detector and Alert System

Chan, Xu Pian (2022) Face Mask Detector and Alert System. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Chan Xu Pian.pdf Restricted to Registered users only Download (4MB)

Abstract

Following the global pandemic of COVID-19, wearing a mask is an important component in preventing COVID-19 transmission. However, some people usually not wearing a mask or wear mask improperly in public places. To solve this problem, the idea of using fine-tuned YOLOv4 to act as face mask detector in real time is proposed in this project. The system is able to detect different types and colors of masks, as well as, classify into three categories which are faces with mask, without mask and wearing the mask incorrect. Custom dataset is created and to be trained and tested. Moreover, the system also comes with an alert system which utilized Selenium in conjunction with Python. The alert system is able to send the captured image and warning message to the admin when anyone against the regulations is detected. Furthermore, a simple counting mechanism also implied in the proposed system to show the count of detections in each category in real time and an excel file will be created when the system is closed. In the end, the model is outperformed the other state-of-art methods by achieving an accuracy of 96.98%, precision of 92.68%, recall score of 94.19%, F1-score of 93.43% and 78.36% of IoU. All the objectives are able to achieve in this project.