ABSTRACT
The World Health Organization (WHO) has publicized a global public health emergency due to the COVID-19 coronavirus pandemic. Wearing a mask in public can provide protection against the spread of disease. Tremendous progress has been made in object detection in recent times, thanks in large part to deep learning models, which have shown encouraging results when it comes to recognizing objects in images. Recent technological developments have made this progress possible. Wearing a mask in public is one way to prevent the transmission of COVID-19 from others. Our study employs You Only Look Once (YOLO) v7 to determine whether a subject is wearing a mask, and then divides them into three groups depending on the degree to which they are wearing a mask correctly (none, bad, and good). In this study, we merged two datasets, the Face Mask Dataset (FMD) and the Medical Mask Dataset (MMD), to conduct our experiment. These models' evaluations and ratings include crucial criteria. According to our data, YOLOv7 achieves the highest mAP (98.5%) in the "Good"class. © 2023 IEEE.
ABSTRACT
With the global outbreak of Corona Virus Disease 2019(COVID-19), many countries had made it mandatory for people to wear masks in public places. This paper proposed a novel mask detection algorithm RMPC (Restructing the Maxpool layer and the Convolution layer)-YOLOv7 based on YOLOv7 for detecting whether people wear masks in public places. The RMPC-YOLOv7 algorithm reconstructed the downsampling structure in the original YOLOv7 algorithm. We changed the stacking of the maxpooling layer and the convolutional layer. This enabled the feature information to be fully integrated to achieve the accuracy improvement of the new model. Through comparison experiments, our proposed RMPC-YOLOv7 had was improved 0.9% and 1.2% for mAP0.5 and mAP0.5:0.95, respectively. The experimental results demonstrated the feasibility of RMPC-YOLOv7. © 2023 IEEE.