Optimization of YOLO v8 Model for Forest Fire Target Detection Based on Coordinate Attention Mechanism

In this study, we introduced a coordinate attention mechanism based on the YOLO v8 model and optimized it to improve the performance of forest fire target detection. We conducted two sets of experiments, the first being the application of YOLO v8 basic model in forest fire target detection. The results showed that the loss values of the training and validation sets gradually decreased and tended to converge, while indicators such as precision, recall, and mean average precision (mAP) steadily improved. In the second experiment, the improved YOLO v8 model also demonstrated good training performance, with continuous reduction in loss values and convergence, as well as significant improvements in accuracy, recall, and mAP metrics. Specifically, compared to the YOLO v8 base model, the improved model improved by 0.03 in accuracy, 0.38 in recall, 0.018 in mAP50, and 0.014 in mAP50-95. This series of data indicates that our proposed optimization scheme effectively enhances the performance of the model in forest fire target detection tasks. This study not only provides a more efficient technical means for forest fire monitoring, but also offers new ideas and methods for optimizing target detection algorithms in related fields, which has important practical application value and theoretical significance.