Realtime Face Mask Detection Using TINY-YOLO V4

Abstract

In the time of the Covid-19 pandemic there is a need to maintain social distancing and prioritize personal hygiene by the use of face masks and proper sanitary precautions. This although is hard to be monitored and controlled accurately and efficiently, can be done through the use of object detection using convolutional neural networks. This can be done in a way using Tiny-YOLOv4 which is an object detection algorithm that provides lightning-fast detection for many classes of objects without the use of such hardware resources. This project aims to train and test a custom data set using this algorithm to create a highly efficient and accurate face mask detection system that can be easily customized to add additional features such as warning systems, etc. It aims to be a system that can prove to be useful once the pandemic is over as it provides crucial data for the prevention and control of any other possible pandemics that may occur in the future.