| 摘 要: 车辆分类与检测在智能交通系统、道路交通规划、安全预警、无人驾驶等领域发挥着越来越重要的作用。随着图形处理器(Graphics Processing Unit, GPU)运算能力的增强以及数据量的剧增，以卷积神经网络为主的深度学习成为研究热点。基于LeNet、AlexNet、VGG、GoogLet、ResNet等卷积神经网络，介绍并分析了6 种车辆分类方法、8 种车辆检测方法及4 个用于评估这些方法的数据集，并对车辆分类及检测的研究方向进行了展望。 |
| 关键词: 卷积神经网络；车辆分类；车辆检测；深度学习 |
|
中图分类号: TP391
文献标识码: A
|
|
| Research on Vehicle Classification and Detection Technology based on Convolutional Neural Network |
|
XIE Dan1, CHEN Lichao2, CAO Lingling1, ZHANG Yanli1
|
( 1.Department of Information Technology Application Innovation and Big Data, Shanxi Jinzhong Institute of Technology, Jinzhong 030600, China ;
2.Department of Big Data, Jinzhong College of Information, Jinzhong 030600, China )
ever_xd@163.com; 2676393759@qq.com; cllzzjzjc0708@163.com; xcjiaowu@163.com
|
| Abstract: Vehicle classification and detection plays an increasingly important role in intelligent transportation systems, road traffic planning, safety warning, unmanned driving and so on. Deep learning based on convolutional neural networks has become a research focus with the enhancement of GPU computing power and the increase in data volume. Based on convolutional neural networks such as LeNet, AlexNet, VGG, GoogLet, ResNet, this paper introduces and analyzes 6 vehicle classification methods, 8 vehicle detection methods and 4 datasets used to evaluate these methods. In addition, the research direction of vehicle classification and detection is prospected in this paper. |
| Keywords: convolutional neural network; vehicle classification; vehicle detection; deep learning |
