无线电工程2025，Vol.55Issue(5)：920-927,8.DOI:10.3969/j.issn.1003-3106.2025.05.003
基于改进YOLOv8的公路车辆火灾检测算法研究
Research on Highway Vehicle Fire Detection Algorithm Based on Improved YOLOv8
摘要
Abstract
To reduce the risk of fire accidents in highway vehicles,reduce casualties,and protect the safety of highway structures,it is particularly important to accurately and quickly detect the occurrence of fires.A fire detection algorithm based on an improved YOLOv8 model is proposed to address the issues of low accuracy and slow detection speed in current highway scenarios.Firstly,redesigning the C2f module and adding an Efficient Channel Attention(ECA)mechanism to improve detection accuracy and reduce interference from vehicle taillights;Then,Shape-Aware Intersection over Union(SIoU)is used to optimize the loss function of original network model,improving regression performance of the bounding box;Finally,a lightweight convolutional Grouped Spatial Convolution(GSConv)is introduced in the Neck module,which enables the model to improve detection speed and ensure detection accuracy,thus enhancing real-time performance of the model.Experimental results show that compared to original model,the improved model performs better on the highway vehicle fire detection dataset,with the mean Average Precision(mAP)increased by 1.8％,the lightweight model parameters decreased by 10％,and the forward transmission time reduced by 13.6％.The improved model has better detection accuracy and speed,and can meet the requirements of real-time fire detection.关键词
目标检测/YOLOv8/高效通道注意力/SIoU/GSConvKey words
object detection/YOLOv8/ECA/SIoU/GSConv分类
交通运输引用本文复制引用
毛紫薇,周正康,唐加山..基于改进YOLOv8的公路车辆火灾检测算法研究[J].无线电工程,2025,55(5):920-927,8.基金项目
南京邮电大学横向科研项目资助(2023外211)Nanjing University of Posts and Telecommunications Horizontal Research Project Grant(2023W211) （2023外211）
