基于人体骨骼信息的列车司机行为智能识别

王志毅; 刘昕武

doi:10.13890/j.issn.1000-128x.2020.04.018

您当前的位置：

首页 >

文章列表页 >

基于人体骨骼信息的列车司机行为智能识别

研究开发 | 更新时间：2021-12-10

- 基于人体骨骼信息的列车司机行为智能识别
- Intelligent Recognition of Train Driver Behavior Based on Human Skeleton Information
- 机车电传动 2020年第4期页码：90-93
- 作者机构：
  
  1.朔黄铁路发展有限责任公司，河北　肃宁　062350
  2.株洲中车时代电气股份有限公司，湖南　株洲　412001
- 作者简介：
  
  王志毅（1970—），男，工程师，主要从事机务运用管理工作。
- 基金信息：
- DOI：10.13890/j.issn.1000-128x.2020.04.018
  中图分类号：
扫描看全文
王志毅, 刘昕武. 基于人体骨骼信息的列车司机行为智能识别[J]. 机车电传动, 2020,(4):90-93.

Zhiyi WANG, Xinwu LIU. Intelligent Recognition of Train Driver Behavior Based on Human Skeleton Information[J]. Electric Drive for Locomotives, 2020,(4):90-93.
王志毅, 刘昕武. 基于人体骨骼信息的列车司机行为智能识别[J]. 机车电传动, 2020,(4):90-93. DOI： 10.13890/j.issn.1000-128x.2020.04.018.

Zhiyi WANG, Xinwu LIU. Intelligent Recognition of Train Driver Behavior Based on Human Skeleton Information[J]. Electric Drive for Locomotives, 2020,(4):90-93. DOI： 10.13890/j.issn.1000-128x.2020.04.018.

摘要

文章提出了一种基于骨架识别的人体姿态捕捉预测算法，用于识别列车驾驶员违章行为。该方法是将司机作业监控视频拆分成一组连续的图片组合，然后通过构建深度学习模型网络预测每一帧图片中的人体姿态，并使用逻辑回归模型对司机行为进行分类，最终，通过将识别出的司机行为与LKJ数据相关联，裁定司机行为是否违章。该方法比目前机务段靠人工抽检的方式更加高效合理，但其识别效果受限于车载摄像头清晰度。

Abstract

A human poseture capture and prediction alcognition was proposed, which was used to identify violated driver. Firstly, driver behavior or action captured by camera was split into a set of successive frames with human poseture. Then each poseture in a frame was recognized by deep learning network and a type of behaviors composed by a set of poseture was classified via logistic regression model. Finally, the association between detected driver behaviors and locomotive LKJ data was applied in the recognition of violated driver. Compared with the daily random-check approach by inspectors in locomotive depots, this way was more reasonable and efficient. However, the accuracy of it was limited by the resolution of surveillance cameras in the driving cab of locomotive.

关键词

人体骨骼深度学习姿态识别LKJ列车

Keywords

human skeletondeep learningposeture recognitionLKJtrain

references

KECKLUND L, INGRE M, KECKLUND G, et al. The train-project: Railway safety and the train driver information environment and work situation-A summary of the main results[J]. Computers in Railways VII, 2000, 50. [2020-06-30]. https://www.witpress.com/elibrary/wit-transactions-on-the-built-environment/50/4182https://www.witpress.com/elibrary/wit-transactions-on-the-built-environment/50/4182. DOI: 10.2495/CR001021http://doi.org/10.2495/CR001021.

MCLEOD R W, WALKER G H, MORAY N. Analysing and modelling train driver performance[J]. Applied Ergonomics, 2005, 36(6): 671-680.

MCLEOD R W, WALKER G H, MORAY N. Assessing the human factors risks in extending the use of AWS[J]. Applied Ergonomics,2005, 36(5): 671-680.

KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]//ACM. Proceedings of the 25th International Conference on Neural Information Processing Systems. New York: Curran Associates Inc., 2012, 1: 1097-1105.

孙志远, 鲁成祥, 史忠植, 等. 深度学习研究与进展[J]. 计算机科学, 2016, 43(2): 1-8.

彭军, 何伟锋, 杨迎泽. 嵌入式列车司机驾驶疲劳检测系统设计[J]. 计算机工程与应用, 2009, 45(33): 57-59.

CAO Z, SIMON T, WEI S E, et al. Realtime multi-person 2D pose estimation using part affinity fields[C]//IEEE. 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 1302-1310. DOI: 10.1109/CVPR.2017.143http://doi.org/10.1109/CVPR.2017.143.

何晓群, 刘文卿. 应用回归分析[M]. 4版. 北京: 中国人民大学出版社, 2015.

WikiMili. Sigmoid function[EB/OL]. 2020-06-21[2020-07-02]. https://wikimili.com/en/Sigmoid_functionhttps://wikimili.com/en/Sigmoid_function.

浏览量

下载量

CSCD

CNKI被引量

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于改进yolov4-tiny的地铁受电弓燃弧检测方法

地铁车辆受电弓异常磨耗研究

基于深度学习的弓网异常状态在线检测研究

基于Mask-RCNN的牵引供电设备红外智能诊断

基于轻量级卷积神经网络的地铁轨道线路状态检测系统