Human Face Mask Detection Using YOLOv7+CBAM in Deep Learning

Abstract

COVID-19 and its variants have affected millions of people around the world. Wearing a mask is an effective way to reduce the spread of the epidemic. While wearing masks is a proven strategy to mitigate the spread, monitoring compliance remains a challenge. In this chapter, the authors propose a mask detection method based on deep learning and convolutional block attention module (CBAM). In this chapter, they extract representative features from input images through supervised learning. In order to improve the recognition accuracy under limited computing resources. They choose YOLOv7 network model and incorporate CBAM into its network structure. Compared with the original version of YOLOv7, the proposed network model improves the mean average precision (mAP) up to 0.3% in face mask detection process. Meanwhile, the method improves the detection speed of each frame 73ms. These advancements have significant implications for real-time, large-scale monitoring systems, thereby contributing to public health and safety.