重庆理工大学学报2025，Vol.39Issue(13)：35-42,8.DOI:10.3969/j.issn.1674-8425(z).2025.07.005
改进YOLOv8的轻量化车辆目标检测算法研究
Research on lightweight vehicle object detection algorithm by improved YOLOv8
摘要
Abstract
To address the issue of complex and dynamic road vehicle conditions and limited detection platform resources,this paper proposes an enhanced vehicle object detection algorithm based on improved YOLOv8.It utilizes an enhanced REPELAN module to replace the C2F module,reducing parameter count and model weight.The original detection head is replaced with a lightweight shared convolutional head(LSCD head),enhancing accuracy while reducing model size.By incorporating a coordinate attention mechanism in the backbone network,the vehicle target detection performance is improved.The WISE-IoU loss function replaces the original CIOU loss function in the YOLOv8 network,accelerating convergence.Experiments employing the processed KITTI and SODA10M datasets demonstrate the improved algorithm increases mAP0.5,mAP0.5～95,recall,and precision by 1.4％,1％,1.8％,and 1.2％compared to YOLOv8n.GFLOP,parameter count,and model volume decrease by 37％,46.7％and 46.8％respectively.It effectively balances model weight reduction and performance,showing excellent generalization capabilities and meeting deployment requirements in resource-constrained environments.关键词
车辆检测/轻量化/YOLOv8/REPVGG/坐标注意力机制/Wise-IoUKey words
vehicle detection/lightweight/YOLOv8/REPVGG/coordinate attention mechanism/wise-IoU分类
信息技术与安全科学引用本文复制引用
蒋康,孙仁云,李卓霖,张国灏,刘金庆..改进YOLOv8的轻量化车辆目标检测算法研究[J].重庆理工大学学报,2025,39(13):35-42,8.基金项目
四川省科技厅重点研发项目(23ZDYF0506) （23ZDYF0506）
成都市科技局重点研发项目(2022-YF05-01047-SN) （2022-YF05-01047-SN）
