Research on Wildfire Smoke Detection Method Based on Improved YOLO11

The detection of wildfire smoke is of significant importance for the prevention of forest fires. In response to the challenges encountered in smoke detection, such as complex backgrounds, small smoke targets, and potential confusion with other elements, this paper proposes a wildfire smoke detection method based on an improved YOLO11. Building upon the latest YOLO11 architecture, we introduce the C2DAmechanism and the DetectAux auxiliary detection head module, improving the detection performance of small-scale smoke targets. Experimental results on a self-constructed wildfire smoke dataset demonstrate that, compared to the original YOLO11 model, the improved model achieves an increase of 1.8% in mean Average Precision (mAP) at IoU thresholds ranging from 0.5 to 0.95, while also exhibiting a faster convergence speed. This method offers more reliable technical support for wildfire early warning systems.