现代电子技术2025，Vol.48Issue(22)：125-132,8.DOI:10.16652/j.issn.1004-373x.2025.22.021
基于YOLOv8-FBFS的危险驾驶行为检测算法
Dangerous driving behavior detection algorithm based on YOLOv8-FBFS
摘要
Abstract
With the continued growth of motor vehicle ownership,traffic safety issues have become increasingly prominent.Driver fatigue and distracted driving behavior are key factors leading to frequent traffic accidents and pose a significant threat to public safety.On this basis,a driving behavior detection algorithm based on improved YOLOv8,YOLOv8-FBFS,is developed for real-time monitoring of driver status and recognition and warning of fatigue and distracted driving behavior.In this algorithm,the SE attention mechanism is introduced to enhance key feature representation while improving detection speed.BiFPN strategy is used to optimize the feature fusion process,ensuring that detection accuracy is not compromised.After comparative experiments and ablation experiments,the improved model can reach 88.9%in accuracy,which is 7.8%higher than that of the original YOLOv8 model,and can reduce 11.1%and 12.9%respectively in model size and calculation volume.These optimizations make the algorithm excellent in real-time detection of fatigue and distracted driving behavior,providing strong technical support for improving road traffic safety.关键词
疲劳驾驶/分心驾驶/YOLOv8/轻量化网络/目标检测/SE注意力机制Key words
fatigue driving/distracted driving/YOLOv8/lightweight networking/target detection/SE attention mechanism分类
信息技术与安全科学引用本文复制引用
何际华,郭佑民,李祯,谷云龙..基于YOLOv8-FBFS的危险驾驶行为检测算法[J].现代电子技术,2025,48(22):125-132,8.基金项目
国家自然科学基金资助项目(72061021) （72061021）
甘肃省自然科学基金资助项目(21JR7RA284) （21JR7RA284）
