Real-Time Detection Method for Wildfire and Smoke Based on an Improved YOLOv11n

To effectively address core challenges in wildfire and smoke detection within complex wildfire environments—including variable target scales, numerous fire-like interferences, and intricate backgrounds—this paper proposes a lightweight real-time detection method based on an improved YOLOv11n. By incorporating the Adaptive Multi-Scale Channel-Spatial Attention (AMS-CSA) module and the Adaptive Path Aggregation Network (APAN), the model enhances cross-scale perception of key discriminative features and optimizes the fusion efficiency of multi-level features. Experimental results demonstrate that the improved model achieves a mAP@0.5 of 78.56%, representing a 2.95% improvement over the baseline YOLOv11n while maintaining a real-time inference speed of 51 FPS. This indicates that the proposed method achieves a favorable balance between detection accuracy and computational efficiency, providing an effective solution for deployment on resource-constrained edge devices to enable early and accurate wildfire warning systems.