Convolution Neural Network Approach for Facial Mask Detection
Irish C. Juanatas, Roben A. Juanatas · 2023
The utilization of Convolutional Neural Network (CNN) in facial mask detection serves as a method to enforce health protocols. In response to the COVID-19 pandemic and the significance of adhering to safety measures, this study employed a You Only Look Once (YOLO) v8 to extract essential characteristics from facial images and categorize into three classes: images with face masks, images without face masks, and images showing improper use of face masks. The vitality of this approach lies in its potential to contribute significantly to public health and safety. By automating the process of mask detection YOLO v8, the model can be deployed in various settings, such as airports, hospitals, workplaces, and public spaces, to identify compliance with mask-wearing guidelines. With a mean average precision (mAP) of 88.2% at 300 epochs, the object detection model demonstrates a positive and excellent performance. The evaluation metrics reveal the model's exceptional precision rate of 96.2%, indicating its ability to confidently and accurately identify positive instances of face masks with minimal false positives. However, the study also acknowledges a recall rate of 73.3%, suggesting that there is room for improvement in capturing all positive instances, and the model may miss some objects during detection.