Face Mask-Wearing Detection Network Based on MSCA-YOLO

Mask detection plays a vital role in public safety and virus prevention. This study proposes MSCA-YOLO, a mask-wearing detection network enhanced by a Multi-Scale Cross-Axis Attention (MSCA) mechanism and dual cross-attention strategy, which improves global feature modeling and semantic extraction. A diverse custom dataset was built under varying natural conditions. Experimental results show that MSCA-YOLO outperforms existing YOLO-based models in Mean Average Precision (mAP) (mAP=89.1) and speed, demonstrating improved robustness and effectiveness.