Research on Smoking Behavior Detection System in Public Places Based on UNS-YOLO Model

This study proposes the UNS-YOLO model for detecting smoking behaviors in public spaces, introducing a new attention mechanism in the backbone network to improve detection accuracy in complex environments. The system identifies smoking behaviors through real-time video data, ensuring efficient monitoring and intervention in public areas. Experimental results demonstrate that the improved model excels in accuracy and recall, and the system has broad application potential for smoking behavior monitoring in public spaces. This work provides significant support for public health initiatives and offers new perspectives for the application of deep learning technologies in this field.