基于深度学习的道路交通标志多目标实时检测

doi:10.16088/j.issn.1001-6600.2020.02.011

摘要/Abstract

摘要： 针对已有道路交通标志检测方法速度慢、受环境影响大、检测效果差等问题,本文提出一种基于Faster-RCNN的道路交通标志多目标实时检测方法。首先,对Faster-RCNN目标检测原理进行深入分析;然后,优化Faster-RCNN网络结构,并选择出合适的预训练模型和网络超参数;最后,在德国交通标志检测数据集(German traffic sign detection dataset, GTSD)上设计多组对比试验,证明了该方法的有效性,单张图片检测时间为0.4 s,准确率达71%以上。在瑞典交通标志检测数据集(Sweden traffic sign detection dataset, STSD)上进行迁移测试,展现了良好的泛化能力,为智能汽车的应用提供了理论依据和技术支持。

关键词: 交通标志, 智能驾驶, 深度学习, Faster-RCNN, 多目标检测

Abstract: To overcome the problems of the existing road traffic sign detection methods, such as slow speed, large environmental impact and poor detection effect, a multi-target real-time detection method for road traffic signs based on Faster-RCNN is proposed. Firstly, the principle of Faster-RCNN target detection is analyzed in depth. Then, the Faster-RCNN network structure is optimized, and the appropriate pre-training model and network hyperparameters are selected. Finally, a set of comparative experiments are designed on the German Traffic Sign Detection dataset (GTSD), which prove the validity of the method. The detection time of single image is 0.4 s, and the accuracy rate is over 71%. The migration test is conducted on the Sweden traffic sign detection dataset (STSD). The method demonstrates a good generalization capability and provides a theoretical basis and technical support for the application of smart cars.

Key words: traffic signs, intelligent driving, deep learning, Faster-RCNN, multi-target detection

中图分类号:

TP391.4

刘英璇, 伍锡如, 雪刚刚. 基于深度学习的道路交通标志多目标实时检测[J]. 广西师范大学学报（自然科学版）, 2020, 38(2): 96-106.

LIU Yingxuan, WU Xiru, XUE Ganggang. Multi-target Real-time Detection for Road Traffic SignsBased on Deep Learning[J]. Journal of Guangxi Normal University(Natural Science Edition), 2020, 38(2): 96-106.

参考文献

[1] SAADNA Y,BEHLOUL A.An overview of traffic sign detection and classification methods[J]. International Journal of Multimedia Information Retrieval,2017,6(3):193-210.DOI:10.1007/s13735- 017-0129-8.
[2] AGHDAM H H,HERAVI E J.Guide to convolutional neural networks[M].Berlin:Springer,2017:1-14.DOI:10.1007/978-3-319-57550-6_1.
[3] REHMAN Y,RIAZ I,FAN Xue,et al.D-patches:effective traffic sign detection with occlusion handling[J].IET Computer Vision,2017,11(5):368-377.DOI:10.1049/iet-cvi.2016.0303.
[4] DIMITRAKOPOULOS G,DEMESTICHAS P.Intelligent transportation systems[J].IEEE Vehicular Technology Magazine,2010,5(1):77-84.DOI:10.1109/MVT.2009.935537.
[5] 王峰,靳小波,于俊伟,等.V-最优直方图及其在车牌分类中的应用研究[J].广西师范大学学报(自然科学版),2013,31(3):138-143.DOI:10.16088/j.issn.1001-6600.2013.03.024.
[6] 余超超,侯进,侯长征.基于显著图与傅里叶描述子的交通标志检测[J].计算机工程,2017,43(5):28-34.DOI:10.3969/j.issn.1000-3428.2017.05.005.
[7] KHAN J F,BHUIYAN S M A,ADHAMI R R.Image segmentation and shape analysis for road-sign detection[J].IEEE Transactions on Intelligent Transportation Systems,2011,12(1):83-96.DOI:10.1109/TITS.2010.2073466.
[8] 李文举,陈奇,董天祯,等.复杂光照条件下交通标志牌检测[J].中国科技论文,2018,13(2):131-135. DOI:10.3969/j.issn.2095-2783.2018.02.003.
[9] BERKAYA S K,GUNDUZ H,OZSEN O,et al.On circular traffic sign detection and recognition[J]. Expert Systems with Applications,2016,48:67-75.DOI:10.1016/j.eswa.2015.11.018.
[10]ELLAHYANI A,El ANSARI M,El JAAFARI I.Traffic sign detection and recognition based on random forests[J].Applied Soft Computing,2016,46:805-815.DOI:10.1016/j.asoc.2015.12.041.
[11]HINTON G E,OSINDERO S,TEH Y W.A fast learning algorithm for deep belief nets[J].Neural Computation,2006,18(7):1527-1554.DOI:10.1162/neco.2006.18.7.1527.
[12]李亚,王广润,王青.基于深度卷积神经网络的跨年龄人脸识别[J].北京邮电大学学报,2017,40(1): 84-88.DOI:10.13190/j.jbupt.2017.01.015.
[13]郝旭政,柴争义.一种改进的深度残差网络行人检测方法[J].计算机应用研究,2019,36(5):1569-1572.DOI:10.19734/j.issn.1001-3695.2017.12.0836.
[14]RAHMANI H,MIAN A,SHAH M.Learning a deep model for human action recognition from novel viewpoints[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2018,40(3):667-681.DOI:10.1109/TPAMI.2017.2691768.
[15]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.DOI:10.1109/TPAMI.2016.2577031.
[16]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Piscataway,NJ:IEEE Press,2014:580-587.DOI:10.1109/CVPR.2014.81.
[17]GIRSHICK R.Fast R-CNN[C]//2015 IEEE International Conference on Computer Vision.Piscataway, NJ:IEEE Press,2015:1440-1448.DOI:10.1109/ICCV.2015.169.
[18]桑军,郭沛,项志立,等.Faster-RCNN 的车型识别分析[J].重庆大学学报,2017,40(7):32-36. DOI:10.11835/j.issn.1000-582X.2017.07.005.
[19]OLMOS R,TABIK S,HERRERA F.Automatic handgun detection alarm in videos using deep learning[J].Neurocomputing,2018,275:66-72.DOI:10.1016/j.neucom.2017.05.012.
[20]ZEILER M D,FERGUS R.Visualizing and understanding convolutional networks[C]//European Conference on Computer Vision.Berlin:Springer,2014:818-833.DOI:10.1007/978-3-319-10590-1_53.
[21]CHATFIELD K,SIMONYAN K,VEDALDI A,et al.Return of the devil in the details:delving deep into convolutional nets[C]//Proceedings of the British Machine Vision Conference.BMVA Press,2014: 1405.3531. DOI:10.5244/C.28.6.
[22]SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition [EB/OL].(2014-09-04)[2019-07-28].https://arxiv.org/abs/1409.1556.
[23]LIU Wei,ANGUELOV D,ERHAN D,et al.SSD:single shot multibox detector[C]//European Conference on Computer Vision.Berlin:Springer,2016:21-37.DOI:10.1007/978-3-319-46448-0_2.
[24]唐聪,凌永顺,郑科栋,等.基于深度学习的多视窗SSD目标检测方法[J].红外与激光工程,2018,47(1):290-298.DOI:10.3788/IRLA201847.0126003.
[25]葛园园,许有疆,赵帅,等.自动驾驶场景下小且密集的交通标志检测[J].智能系统学报,2018,13(3):366-372.DOI:10.11992/tis.201706040.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed