计算机应用 ›› 2019, Vol. 39 ›› Issue (7): 1894-1898.DOI: 10.11772/j.issn.1001-9081.2018122448

• 人工智能 • 上一篇    下一篇

基于长短期记忆的车辆行为动态识别网络

卫星1,2, 乐越1, 韩江洪1,2, 陆阳1,2   

  1. 1. 合肥工业大学 计算机与信息学院, 合肥 230601;
    2. 安全关键工业测控技术教育部工程研究中心(合肥工业大学), 合肥 230601
  • 收稿日期:2018-12-11 修回日期:2019-02-04 出版日期:2019-07-10 发布日期:2019-03-28
  • 通讯作者: 乐越
  • 作者简介:卫星(1980-),男,安徽合肥人,副教授,博士,CCF会员,主要研究方向:物联网工程、离散事件动态系统;乐越(1994-),男,安徽宣城人,硕士研究生,CCF会员,主要研究方向:分布式控制;韩江洪(1954-),男,安徽庐江人,教授,博士生导师,硕士,主要研究方向:计算机网络与通信、计算机控制、汽车电子;陆阳(1967-),男,安徽合肥人,教授,博士生导师,博士,CCF高级会员,主要研究方向:分布式控制、无线通信网络、工业物联网。
  • 基金资助:

    国家重点研发计划专项(2018YFC0604404)。

Vehicle behavior dynamic recognition network based on long short-term memory

WEI Xing<sup>1,2</sup>, LE Yue<sup>1</sup>, HAN Jianghong<sup>1,2</sup>, LU Yang<sup>1,2</sup>   

  1. 1. School of Computer Science and Information Engineering, Hefei University of Technology, Hefei Anhui 230601, China;
    2. Engineering Research Center of Safety Critical Industry Measure and Control Technology, Ministry of Education(Hefei University of Technology), Hefei Anhui 230601, China
  • Received:2018-12-11 Revised:2019-02-04 Online:2019-07-10 Published:2019-03-28
  • Supported by:

    This work is partially supported by the National Key Research Development Program of China (2018YFC0604404).

摘要:

高级辅助驾驶装置采用机器视觉技术实时处理摄录的行车前方车辆视频,动态识别并预估其姿态和行为。针对该类识别算法精度低、延迟大的问题,提出一种基于长短期记忆(LSTM)的车辆行为动态识别深度学习算法。首先,提取车辆行为视频中的关键帧;其次,引入双卷积网络并行对关键帧的特征信息进行分析,再利用LSTM网络对提取出的特性信息进行序列建模;最后,通过输出的预测得分判断出车辆行为类别。实验结果表明,所提算法识别准确率可达95.6%,对于单个视频的识别时间只要1.72 s;基于自建数据集,改进的双卷积算法相比普通卷积网络在准确率上提高8.02%,与传统车辆行为识别算法相比准确率提高6.36%。

关键词: 车辆行为, 长短期记忆网络, 高级辅助驾驶, 深度学习, 卷积神经网络

Abstract:

In the advanced assisted driving device, machine vision technology was used to process the video of vehicles in front in real time to dynamically recognize and predict the posture and behavior of vehicle. Concerning low precision and large delay of this kind of recognition algorithm, a deep learning algorithm for vehicle behavior dynamic recognition based on Long Short-Term Memory (LSTM) was proposed. Firstly, the key frames in vehicle behavior video were extracted. Secondly, a dual convolutional network was introduced to analyze the feature information of key frames in parallel, and then LSTM network was used to sequence the extracted characteristic information. Finally, the output predicted score was used to determine the behavior type of vehicle. The experimental results show that the proposed algorithm has an accuracy of 95.6%, and the recognition time of a single video is only 1.72 s. The improved dual convolutional network algorithm improves the accuracy by 8.02% compared with ordinary convolutional network and increases by 6.36% compared with traditional vehicle behavior recognition algorithm based on a self-built dataset.

Key words: vehicle behavior, Long Short-Term Memory (LSTM) network, advanced assisted driving, deep learning, Convolutional Neural Network (CNN)

中图分类号: