摘要
为了打击假牌、套牌车及以汽车为作案工具的犯罪,且由于传统单一的车型或颜色识别已显得力不从心,因此,提出了改进的多标签深度学习车型与颜色识别模型。该模型利用卷积神经网络自主学习有用特征,利用小卷积核构建深层网络提升模型对复杂函数的表达能力,以全局平均池化取代部分全连接层,减少参数与模型所占空间内存;并利用"单模型多标签"特性将车型与颜色信息融合,使提取到的特征表现力更强。在自建数据集下的实验结果表明,该模型能获得较好的识别结果和较高的准确率,特别是对相同子品牌的不同年款的大规模车型和颜色识别效果更佳,在刑侦稽查时能有效缩小搜索范围并迅速锁定类似目标车辆信息。
In order to combat false license plate vehicle and fake license plate vehicle and crimes of taking avehicle as a tool, and because the traditional recognition model with single vehicle type or color is powerless, sothe vehicle type and color recognition based on improved depth learning with multi-label is proposed. The modeluses convolutional neural network to learn the useful features autonomously, and uses the small convolutionkernel to construct a deep network to enhance the model expression ability to express complex functions. Then,the global average pooling is used to replace the partial fully connected layers to reduce the parameters and themodel memory space. By using the ^single model with multi-labeP5 to combine vehicle type information withcolor information, the extracted features is more expressive. The experimental results on dataset show that themodel can obtain better recognition results and higher accuracy, especially for large-scale vehicle type and colorrecognition of different years and styles from the same vehicle sub-brand. Therefore, it can narrow the searchrange effectively and lock the similar target vehicle information quickly in the criminal investigation.
作者
赵珊
黄强强
曲宏山
刘相利
ZHAO Shan;HUANG Qiang-qiang;QU Hong-shan;LIU Xiang-li(School of Computer Science and Technology, Henan Polytechnic University, Jiaozuo 454000, China;School of Computer Science, Henan University of Engineering, Zhengzhou 45 119 1, China)
出处
《测控技术》
CSCD
2018年第2期3-6,10,共5页
Measurement & Control Technology
基金
河南省基础与前沿技术研究资助项目(132300410462)
关键词
车型识别
颜色识别
多标签深度学习
卷积神经网络
智能交通系统
vehicle recognition
color recognition
multi-label depth learning
convolutional neural network
intelligenttransportation system
