摘要
现有车辆识别算法模型在特征提取方面可能不够精准,无法有效提取车辆的关键特征,模型的泛化能力有限,在面对新的、未见过的车辆时,识别效率会大打折扣。因此,本文探讨改进YOLOv5算法在提升车辆图像识别效率方面的应用。通过在YOLOv5算法中融入注意力机制和知识蒸馏技术,有效提升了模型的车辆特征提取能力以及模型泛化能力。实验结果表明,改进后的算法关键指标均有显著提升,为智能交通系统中的车辆检测提供了更高效准确的解决方案。
The existing vehicle recognition algorithm models may not be accurate enough in feature extraction,and cannot effectively extract key features of vehicles.The generalization ability of the model is limited,and the recognition efficiency will be greatly reduced when facing new and unseen vehicles.Therefore,this article explores the application of improving the YOLOv5 algorithm in enhancing the efficiency of vehicle image recognition.By incorporating attention mechanisms and knowledge distillation techniques into the YOLOv5 algorithm,the model’s ability to extract vehicle features and generalize has been effectively improved.The experimental results show that the key indicators of the improved algorithm have significantly improved,providing a more efficient and accurate solution for vehicle detection in intelligent transportation systems.
作者
杨小琴
Syazwina Alias
王慧
宁振国
YANG Xiaoqin;Syazwina Alias;WANG Hui;NING Zhenguo(Nanjing Tech University Pujiang Institute,Nanjing Jiangsu 210034;Segi University,Selangor Malaysia 47810)
出处
《软件》
2024年第11期13-16,共4页
Software
基金
教育部产学合作协同育人项目(22060091906233)
全国高等院校计算机基础教育教学研究资助专项(2023-AFCEC-042)
南京工业大学浦江学院计算机与通信工程学项目(2022XYL03)。
关键词
改进算法
YOLOv5
车辆图像
识别效率
improved algorithm
YOLOv5
vehicle image
identification efficiency
