Exploring the potential of YOLOv8 in hybrid models for facial mask identification in diverse environments

The use of facial masks has become a global standard in response to the COVID-19 pandemic, necessitating the development of reliable detection systems to monitor adherence to health guidelines. This paper presents an facial mask detection framework that enhances the YOLOv8 algorithm and leverages the ResNet-50 architecture for improved feature extraction. Our proposed framework was rigorously evaluated against widely recognized methods, including the original YOLOv8, faster R-CNN, and SSD, across various metrics such as precision, recall, and mean average precision (mAP). The results of these comparative analyses underscore the superiority of our approach, with our framework showing significant improvements in the detection of properly worn masks. Specifically, our model achieved precision and recall rates above 97% and a mAP0.5% of 99.6%, significantly outperforming the comparison group. These advancements highlight the potential of our framework as a cornerstone for public health initiatives, providing a high-precision tool for real-time mask detection and adherence monitoring in various settings.