Mask wearing detection in public places based on YOLOv5
Jinqing Liu, Chao Zhang, Boya Ma, Zixu Wang, Shuai Lei · 2021 3rd International Conference on Machine Learning, Big Data and Business Intelligence (MLBDBI) · 2021
The new type of coronary pneumonia has not completely ended. Wearing a mask in public is not only the most simple and effective means of epidemic prevention, but also an important epidemic prevention policy. In order to efficiently, accurately and robustly detect whether people wear masks in public scenes, firstly, the mask data set is established, and the self-made crowd image data set is trained based on YOLOv5s model in Google Colaboratory, so as to obtain the optimal detection model. Finally, the verification experiment of the detection model is carried out in the actual scene. The experimental results show that the detection accuracy of personnel wearing masks based on YOLOv5s detection model is 87%, and it is easy to deploy in low-performance mobile devices.