中文核心期刊
CSCD来源期刊
中国科技核心期刊
RCCSE中国核心学术期刊

重庆交通大学学报(自然科学版) ›› 2020, Vol. 39 ›› Issue (01): 8-14.DOI: 10.3969/j.issn.1674-0696.2020.01.02

• 交通+大数据人工智能 • 上一篇    下一篇

基于卷积神经网络的汽车试验场外物入侵识别

向华荣1,2,曾敬1,2   

  1. (1. 重庆西部汽车试验场管理有限公司,重庆 408300; 2. 中国汽车工程研究院股份有限公司,重庆 401122)
  • 收稿日期:2018-04-17 修回日期:2019-11-22 出版日期:2020-01-12 发布日期:2020-01-12
  • 作者简介:向华荣(1979—), 男, 重庆人, 高级工程师,硕士,主要从事汽车试验技术方面的研究。E-mail:xianghuarong@cxapg.com。
  • 基金资助:
    工业强基工程项目 (0714-EMTC02-5593/20)

Recognition on Invaders into Automobile Proving Ground Based on Convolution Neural Network

XIANG Huarong1,2,ZENG Jing1,2   

  1. (1.Chongqing Xibu Automotive Proving Ground Management Co., Ltd., Chongqing 408300, China; 2. China Automotive Engineering Research Institute Co., Ltd., Chongqing 401122, China)
  • Received:2018-04-17 Revised:2019-11-22 Online:2020-01-12 Published:2020-01-12

摘要: 为了对汽车试验场的外物入侵进行识别并预警,结合现有监控系统,利用Tensorflow搭建Faster-RCNN框架,分别使用InceptionV2和ResNet101作为核心卷积神经网络,用自建的标注数据集进行最终训练,利用得到的模型进行迁移测试,试验结果表明:InceptionV2系统的平均精度值为81.7%,ResNet101系统的平均精度值为84.1%。将两种系统结合Opencv的图像抓取功能及现有摄像监控设备进行联合测试,结果表明两种系统均能在阴天和低像素摄像头搭配下实时对高速或慢速移动的物体进行识别、分类、标注、预警。

关键词: 车辆工程, 汽车试验场, 外物入侵, 卷积神经网络, 目标检测

Abstract: In order to recognize and alarm invaders which entered into automobile proving ground, Faster-RCNN framework was established by Tensorflow, combining with current video monitoring system. InceptionV2 and ResNet101 were taken as core convolution neural network respectively. The final training was carried out with the self-built annotation data set, and the migration test was carried out with the obtained model. The experimental results show that the average precision value of InceptionV2 system and ResNet101 system reaches 81.7% and 84.1% respectively. The joint test combining the two systems with the image capture function of Opencv and current video monitoring system was carried out. The results prove that both systems can recognize, classify, mark and alarm the high-speed or low-speed moving objects in real time under cloudy weather or low-pixel camera condition.

Key words: vehicle engineering, automobile proving ground, invaders, convolution neural network, target detection

中图分类号: