基于深度学习的智轨信号灯检测和识别方法

熊群芳; 林军; 袁希文; 徐阳翰; 岳伟; 李源征宇

doi:10.13890/j.issn.1000-128X.2024.01.244

您当前的位置：

首页 >

文章列表页 >

基于深度学习的智轨信号灯检测和识别方法

智能化技术 | 更新时间：2025-02-08

- 基于深度学习的智轨信号灯检测和识别方法
- Traffic light detection and recognition based on deep learning for autonomous-rail rapid tram
- 机车电传动 2024年第6期页码：140-148
- 作者机构：
  
  中车株洲电力机车研究所有限公司，湖南株洲 412001
- 作者简介：
  
  熊群芳，女，硕士，图像处理工程师，主要从事图像处理方面的研究；E-mail: xiongqf@csrzic.com
- 基金信息：
  
  国家重点研制计划项目(2022YFB4300405)
- DOI：10.13890/j.issn.1000-128X.2024.01.244
  中图分类号： U239.5;TP391
- 纸质出版日期：2024-11-10，
  
  收稿日期：2023-07-10，
  
  修回日期：2024-04-15，
- 稿件说明：
移动端阅览
熊群芳, 林军, 袁希文, 等. 基于深度学习的智轨信号灯检测和识别方法[J]. 机车电传动, 2024(6): 140-148.

XIONG Qunfang, LIN Jun, YUAN Xiwen, et al. Traffic light detection and recognition based on deep learning for autonomous-rail rapid tram[J]. Electric drive for locomotives,2024(6): 140-148.
熊群芳, 林军, 袁希文, 等. 基于深度学习的智轨信号灯检测和识别方法[J]. 机车电传动, 2024(6): 140-148. DOI：10.13890/j.issn.1000-128X.2024.01.244.

XIONG Qunfang, LIN Jun, YUAN Xiwen, et al. Traffic light detection and recognition based on deep learning for autonomous-rail rapid tram[J]. Electric drive for locomotives,2024(6): 140-148. DOI：10.13890/j.issn.1000-128X.2024.01.244.

摘要

智轨电车作为中车株洲电力机车研究所有限公司自主研发的智能轨道快运系统（简称“智轨”），其交通信号灯的检测与识别是提升智轨自动驾驶系统安全性的关键技术。智轨交通信号灯除了少部分通用信号灯，绝大多数为定制信号灯，而目前已有的信号灯检测与识别方法无法满足智轨自动驾驶环境下的检测要求。因此，文章利用深度学习算法对智轨信号灯检测与识别开展了相关研究工作，首先，通过高精度地图信息确定信号灯的RoI( Region of Interest )区域，缩小对智轨信号灯检测的范围，提升检测速度；其次，采用改进的YOLOV5s网络对RoI区域进行特征提取，检测出智轨交通信号灯；最后，对提取的交通信号灯图片采用MobileNetV2轻量级网络进行识别分类，确定信号灯的具体类别。为进一步增强模型的泛化性能，在信号灯检测之前增加了图像诊断算法，针对曝光、逆光等复杂环境及时提示预警，同时保存这些非正常数据，用于信号灯的检测与分类模型训练，进一步优化模型。试验结果表明，文章提出的方法对于智轨信号灯的检测与识别取得较好效果，白天在指定道路上平均精度均值达到84.76%，且实时性能良好。

Abstract

The autonomous rail rapid transit (ART) system is an intelligent express transport solution developed independently by CRRC Zhuzhou institute co.

ltd. The detection and recognition of traffic lights are critical for improving the safety of the ART automatic operation system. However

existing means for traffic light detection and recognition often fall short of meeting the detection requirements in the automatic operation environments characterized by customized traffic lights

with only a few regular signals. This paper presents study efforts in this field through the application of a deep learning algorithm. Firstly

regions of interest (RoIs) for traffic lights were determined using high-precision map information to narrow the detection range and improve the detection speed. Secondly

an improved YOLOV5s network was employed to extract features from the RoIs

facilitating the recognition of ART traffic lights. Finally

the extracted traffic signal images were classified using a MobileNetV2 lightweight network to identify specific signal categories. In order to further enhance the model's generalization performance

an image diagnosis algorithm was introduced before signal recognition

which generates warnings for complex conditions

such as overexposure and backlighting. These abnormal image data were saved and utilized for further training and optimizing the model. Experimental results reveals that the proposed approach effectively detected and recognized ART traffic lights

achieving an average detection precision of 84.76% on designated roads during the daytime while exhibiting good real-time performance.

关键词

深度学习智轨交通信号灯高精度地图YOLOV5sMobileNetV2图像诊断

Keywords

deep learningART traffic lighthigh-precision mapYOLOV5sMobileNetV2image diagnosis

references

OMACHI M, OMACHI S. Detection of traffic light using structural information[C]//IEEE. IEEE 10th International Conference on Signal Processing Proceedings. Beijing: IEEE, 2010: 809-812.

许明文, 张重阳. 基于显著性特征的交通信号灯检测和识别[J]. 计算机与数字工程, 2017, 45(7): 1397-1401.

XU Mingwen, ZHANG Chongyang. Traffic light detection and recognition based on salience map[J]. Computer and digital engineering, 2017, 45(7): 1397-1401.

KIM Y K, KIM K W, YANG Xiaoli. Real time traffic light recognition system for color vision deficiencies[C]//IEEE. 2007 International Conference on Mechatronics and Automation. Harbin: IEEE, 2007: 76-81.

OMACHI M, OMACHI S. Traffic light detection with color and edge information[C]//IEEE. 2009 2nd IEEE International Conference on Computer Science and Information Technology. Beijing: IEEE, 2009: 284-287.

KASTNER R, MICHALKE T, BURBACH T, et al. Attention-based traffic sign recognition with an array of weak classifiers[C]//IEEE. 2010 IEEE Intelligent Vehicles Symposium. La Jolla: IEEE, 2010: 333-339.

贾瑞明, 刘立强. 基于卷积神经网络的交通灯识别[J]. 北方工业大学学报, 2018, 30(5): 22-30.

JIA Ruiming, LIU Liqiang. Traffic light recognition based on convolutional neural network[J]. Journal of north China university of technology, 2018, 30(5): 22-30.

LEE E, KIM D. Accurate traffic light detection using deep neural network with focal regression loss[J]. Image and vision computing, 2019, 87: 24-36.

李昌兴, 黄艳虎, 支晓斌, 等. 基于加速k均值的谱聚类图像分割算法改进[J]. 传感器与微系统, 2016, 35(9): 137-140.

LI Changxing, HUANG Yanhu, ZHI Xiaobin, et al. Improvements of acceleration k-means based spectral clustering algorithm for image segmentation[J]. Transducer and microsystem technologies, 2016, 35(9): 137-140.

LU Xiangning, HE Zhenzhi, SU Lei, et al. Detection of micro solder balls using active thermography technology and K-means algorithm[J]. IEEE transactions on industrial informatics, 2018, 14(12): 5620-5628.

GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//IEEE. 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus: IEEE, 2014: 580-587.

REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE transactions on pattern analysis and machine intelligence, 2017, 39(6): 1137-1149.

DAI Jifeng, LI Yi, HE Kaiming, et al. R-FCN: object detection via region-based fully convolutional networks[C]//ACM. Proceedings of the 30th International Conference on Neural Information Processing Systems. Barcelona: Curran Associates Inc., 2016: 379-387.

陈飘依. 基于深度模型的实时交通灯检测[D]. 北京: 北京邮电大学, 2019.

CHEN Piaoyi. Real-time traffic-light detection on based on depth model[D]. Beijing: Beijing University of Posts and Telecommunications, 2019.

REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. (2018-04-08) [2024-05-16]. https://arxiv.org/abs/1804.02767https://arxiv.org/abs/1804.02767.

LIU Wei, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]//Springer. Computer Vision – ECCV 2016. Cham: Springer, 2016: 21-37.

谭思奇. 基于深度学习的道路交通信号灯检测与识别方法研究[D]. 重庆: 重庆交通大学, 2022.

TAN Siqi. Research on road traffic light detection and recognition method based on deep learning[D]. Chongqing: Chongqing Jiaotong University, 2022.

邓天民, 谭思奇, 蒲龙忠. 基于改进YOLOv5s的交通信号灯识别方法[J]. 计算机工程, 2022, 48(9): 55-62.

DENG Tianmin, TAN Siqi, PU Longzhong. Traffic light recognition method based on improved YOLOv5s[J]. Computer engineering, 2022, 48(9): 55-62.

熊辉, 郭宇昂, 陈超义, 等. 基于遗传优化与深度学习的交通信号灯检测[J]. 汽车工程, 2019, 41(8): 960-966.

XIONG Hui, GUO Yuang, CHEN Chaoyi, et al. Traffic light detection based on genetic optimization and deep learning[J]. Automotive engineering, 2019, 41(8): 960-966.

BEHRENDT K, NOVAK L, BOTROS R. A deep learning approach to traffic lights: detection, tracking, and classification[C]//IEEE. 2017 IEEE International Conference on Robotics and Automation. Singapore: IEEE, 2017: 1370-1377.

SANDLER M, HOWARD A, ZHU Menglong, et al. Inverted residuals and linear bottlenecks: mobile networks for classification, detection and segmentation[EB/OL]. (2018-01-16) [2024-05-16]. https://arxiv.org/abs/1801.04381v2https://arxiv.org/abs/1801.04381v2.

王艺皓, 丁洪伟, 李波, 等. 复杂场景下基于改进YOLOv3的口罩佩戴检测算法[J]. 计算机工程, 2020, 46(11): 12-22.

WANG Yihao, DING Hongwei, LI Bo, et al. Mask wearing detection algorithm based on improved YOLOv3 in complex scenes[J]. Computer engineering, 2020, 46(11): 12-22.

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

基于三维点云与图像融合的轨道交通场景行人检测方法

基于低秩分解的YOLO轻量化目标检测模型

轨道交通车载端到端语音合成

基于改进yolov4-tiny的地铁受电弓燃弧检测方法

基于深度学习的弓网异常状态在线检测研究