基于 MobileNetV3 与 ST-SRU 的危险驾驶行为识别研究
作者:
作者单位:

杭州电子科技大学自动化学院

作者简介:

通讯作者:

中图分类号:

TP391.4;TP18

基金项目:

国家自然科学基金项目(面上项目,重点项目,重大项目)


Research on dangerous driving behaviors recognition based on MobileNetV3 and ST-SRU
Author:
Affiliation:

School of Automation, Hangzhou Dianzi University

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对危险驾驶行为引起的交通安全事故频发的现状,本文提出一种基于MobileNetV3和ST-SRU的危险驾驶行为识别系统。首先,修改MobileNetV3的网络结构使其用于人体姿态估计,输出关节点的热力图和偏移量图,用来估计 个关节点的坐标位置;其次,定义了ST-SRU动作识别算法,利用动作的骨架序列数据,对动作进行分类。实验结果表明,MobileNetV3姿态估计算法在自建的AI Challenger上肢姿态数据集上测得PCP值(Percentage Correct Parts)达到95.6%,测试1000次速度仅为5.03秒。利用自建的危险驾驶行为数据集,将训练好的姿态估计和动作识别模型移植到嵌入式平台,实现了实时的危险驾驶行为识别系统。

    Abstract:

    In the face of frequent traffic accidents caused by dangerous driving behaviors, this paper proposes a dangerous driving behavior recognition system based on MobileNetV3 and ST-SRU. Firstly, the network structure of MobileNetV3 is modified to be used for human posture estimation, and the heatmaps and offsets of joint points are output to estimate the coordinate positions of joint points; Secondly, the ST-SRU action recognition algorithm is defined, and the actions are classified by using skeleton sequence data. The experimental results show that the PCP (percentage correct parts) of MobileNetV3 pose estimation algorithm is 95.6% on the self-built AI Challenger upper limb attitude dataset, and the speed of 1000 tests is only 5.03 seconds. By using the self-built dangerous driving behavior dataset, the trained pose estimation and action recognition model is transplanted to the embedded platform, and the real-time dangerous driving behavior recognition system is realized.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2020-08-17
  • 最后修改日期:2021-02-28
  • 录用日期:2021-03-03
  • 在线发布日期: 2021-04-01
  • 出版日期: