Fire and Smoke Detection Method for Aging Communities Based on Improved YOLOv8n

To effectively deal with the special challenges of fire warning in aging communities and ensure the safety of the living environment of the elderly, a fire and smoke detection method based on improved YOLOv8n is proposed. By using MobileNetV3-Small as the backbone, combined with the lightweight hybrid attention mechanism Shuffle Attention (SA) and a-EIoU loss function, the model's detection performance on low-computing power devices is improved. Experimental results show that the mAP of the improved YOLOv8n reaches 66.77%, which is 3.74% higher than the baseline model. At the same time, the computational cost is significantly reduced, FLOPs and Param are reduced by 42% and 37.9% respectively, and the FPS reaches 77. This shows that the improved YOLOv8n not only performs well in detection accuracy, but also has significant advantages in computational efficiency and real-time processing. It is very suitable for resource-constrained aging community environments and provides effective protection for the safety of life and property of the elderly.