Akademik Veri Yönetim Sistemi
2025 Innovations in Intelligent Systems and Applications Conference, ASYU 2025, Bursa, Türkiye, 10 - 12 Eylül 2025, (Tam Metin Bildiri)
During the COVID-19 pandemic, the use of face masks became mandatory in most countries and played a critical role in reducing transmission. Even in the postpandemic era, proper mask usage remains important in many sectors, including healthcare, transportation, and high-density workplaces. This study presents a lightweight and accurate deep learning approach for detecting the condition of face mask usage specifically, whether a person is wearing a mask correctly, incorrectly, or not at all. The YOLOv8s object detection model was fine-tuned from COCO-pretrained weights on a publicly available dataset of 853 annotated images. Proposed model achieved a mean average precision of 89.4%, a recall of 76.9%, and a precision of 92.9%. These results demonstrate superior performance compared to earlier approaches like YOLOv3-tiny and Mask R-CNN on the same dataset, showing the potential of this method for real-world mask monitoring in safety-critical environments.