基于改进YOLOv8的火灾目标检测系统
Fire Target Detection System Based on Improved YOLOv8

火灾发生初期，烟雾状态变化多端火焰的尺寸大小也非常小，现有的目标检测算法面对这复杂情况下会出现检测速度慢、检测准确率低。针对类似这样的问题，本文提出了基于改进YOLOv8的火灾目标检测系统。在YOLOv8的骨干网络末端添加BotNet结构，用来增强网络对火灾的特征提取，在YOLOv8的头部末端引入EMA注意力机制防止权重剧烈变化。改进的YOLOv8模型提高了目标检测的精确度。实验的结果表明，改进的YOLOv8模型与YOLOv8模型对比，改进的YOLOv8模型在mAP上提高了2.3%、火灾与烟雾的预测准确率也分别提高了1.4%和1%，进一步说明改进的YOLOv8模型可以满足对火灾的目标检测。
In the early stages of a fire, the smoke state changes frequently and the size of the flame is also very small. Faced with this complex situation, existing target detection algorithms will have slow detection speed and low detection accuracy. In response to problems like this, this article proposes a fire target detection system based on improved YOLOv8. The BotNet structure is added to the end of the backbone network of YOLOv8 to enhance the network’s feature extraction of fires, and the EMA attention mechanism is introduced at the head end of YOLOv8 to prevent drastic changes in weights. The improved YOLOv8 model improves the accuracy of target detection. The results of the experiment show that compared with the YOLOv8 model, the improved YOLOv8 model increased mAP by 2.3%, and the fire and smoke prediction accuracy also increased by 1.4% and 1% respectively, further demonstrating that the improved YOLOv8 model can Meet the target detection of fire.