计算机应用与软件2017，Vol.34Issue(11)：228-231,4.DOI:10.3969/j.issn.1000-386x.2017.11.042
基于卷积神经网络的汽车型号识别
VEHICLE MODEL RECOGNITION BASED ON CONVOLUTIONAL NEURAL NETWORK
陈淑君 1周永霞 1方勇军2
作者信息
- 1. 中国计量学院信息工程学院 浙江杭州310018
- 2. 杭州吾思智能科技有限公司 浙江杭州310018
- 折叠
摘要
Abstract
Using the front view of the vehicle,we propose an algorithm for vehicle model recognition based on deep learning.First,the vehicle face region was detected by using the cascade detector of Adaboost algorithm.Then,a convolution neural network was designed to identify the region.Moreover,we compared the recognition effects of SVM,ELM classifier and popular manual design features (SIFT,HOG and LBP) respectively.The experimental results show that the recognition rate of the deep learning is significantly higher than the traditional machine learning method.Deep learning demonstrates excellent performance.关键词
深度学习/卷积神经网络/Adaboost/LBP/SVM/车辆识别Key words
Deep learning/Convolution neural network/Adaboost/LBP/SVM/Vehicle identification分类
信息技术与安全科学引用本文复制引用
陈淑君,周永霞,方勇军..基于卷积神经网络的汽车型号识别[J].计算机应用与软件,2017,34(11):228-231,4.
