Research on Smoke and Fire Detection Algorithm Based on YOLOv11

Currently, the fire rescue field is still facing many technical challenges, in which the accurate detection of fire source smoke and real-time response in complex environments is the core problem to be solved. Traditional detection methods mainly rely on sensor technology or threshold segmentation algorithms, which are susceptible to interference by environmental factors, and it is difficult to achieve an effective balance between detection speed and accuracy. Although deep learning techniques have made significant progress in the field of fire detection, there are still obvious limitations in areas such as detection in complex backgrounds and multi-target scenes. Aiming at the above technical bottlenecks, this study uses a YOLOv11-based fire-smoke target detection algorithm. By using the YOLOv11 model, the feature extraction ability of irregular targets such as smoke and flame is significantly enhanced; meanwhile, the detection robustness in complex scenes is effectively improved due to the loss function and multi-scale detection architecture. The experimental results show that the algorithm achieves 77.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> mean average precision (mAP) and 82.9% accuracy on the complex fire source smoke dataset, which provides a reliable technical support for fire warning and rescue command.