自然场景下的中国交通标志检测算法

doi:10.19678/j.issn.1000-3428.0068938

摘要/Abstract

摘要：

当前在自然场景下对中国交通标志进行检测时通常存在检测精度和检测速度不平衡的问题, 为此, 提出一种基于YOLOv5的改进算法。首先根据ShuffleNet V2网络的设计理念, 提出改进型轻量化卷积块来代替YOLOv5中的卷积块, 以减少网络整体的计算量和参数量; 其次提出一个扩大感受野的轻量化模块ASPC来代替原网络中的空间金字塔池化模块SPP, 从而降低网络计算量和参数量同时提升网络的检测精度; 最后把颈部特征融合网络中的上采样模块替换为CARAFE上采样算子, 并提出多尺度通道混洗注意力机制MCSA, 将其添加在CARAFE算子之后, 让网络融合全局与局部的特征信息, 更有效地减少颈部特征融合网络对交通标志特征信息的丢失。在自制的中国多类交通标志数据集CMTSD上进行实验, 结果表明, 改进后的算法模型大小相较于原模型减少了41%, 每秒检测帧数(FPS)提高了9.37, 平均检测识别精度mAP@0.5提升了2.91%, 达到94.76%。改进算法在不同的自然场景下均能达到较好的检测效果, 可以满足实际场景中对交通标志检测的需求。

关键词: 深度学习, 交通标志检测, YOLOv5网络, 轻量化网络, 全局与局部融合

Abstract:

This study proposes an improved YOLOv5 traffic sign detection algorithm to address detection accuracy and speed imbalance problems for Chinese traffic sign detection in natural scenes. First, according to the ShuffleNet V2 network design concept, an improved lightweight convolutional block is proposed as a substitute for the YOLOv5 convolutional block. This improvement aims to decrease both the computational load and number of network parameters. Second, a lightweight module, the Atrous Spatial Pyramid Convolution (ASPC), is introduced to enlarge the receptive field, to replace the Spatial Pyramid Pooling (SPP) module in the original network. This decreases the computational and parameter complexity of the network and enhances its detection accuracy. Finally, the upsampling module in the neck is replaced by Content-Aware ReAssembly of FEatures (CARAFE), and a Multi-scale Channel Shuffle Attention (MCSA) mechanism is incorporated next to the CARAFE. This modification enables the network to effectively fuse global and local feature information, thereby more efficiently reducing the loss of traffic sign feature information in the neck. Experimental results on the Chinese Multi-class Traffic Sign Dataset(CMTSD)demonstrate that the optimized algorithm model achieves a significant size reduction of 41% compared with YOLOv5. Concurrently, a notable detection speed increase of 9.37 Frames Per Second (FPS) is observed. Moreover, the mAP@0.5 exhibits a substantial increase of 2.91%, reaching a remarkable value of 94.76%. This enhanced performance enables the algorithm to deliver excellent detection results spanning various natural scene environments and effectively meets traffic sign detection requirements for real-world scenarios.

Key words: deep learning, traffic sign detection, YOLOv5 network, lightweight network, global and local fusion

王翰文, 葛青, 朱宁可, 余鹏飞. 自然场景下的中国交通标志检测算法[J]. 计算机工程, 2024, 50(11): 327-337.

WANG Hanwen, GE Qing, ZHU Ningke, YU Pengfei. Chinese Traffic Sign Detection Algorithm in Natural Scenes[J]. Computer Engineering, 2024, 50(11): 327-337.

https://www.ecice06.com/CN/Y2024/V50/I11/327

图/表 13

图1 YOLOv5网络结构

Fig.1 YOLOv5 network structure

图2 MSC3结构

Fig.2 MSC3 structure

图3 ASPC结构

Fig.3 ASPC structure

图4 CARAFE结构

Fig.4 CARAFE structure

图5 MCSA结构

Fig.5 MCSA structure

图6 改进的YOLOv5网络结构

Fig.6 Improved YOLOv5 network structure

图7 自制CMTSD数据集中的具体类别划分

Fig.7 Specific category classification in self-made CMTSD dataset

图8 自制CMTSD数据集图片示例

Fig.8 Example of self-made CMTSD dataset images

图9 2种算法检测结果对比

Fig.9 Comparison of the detection results of two algorithms

参考文献 31

1	陈进. 浅析中国城市智能交通系统产业化发展趋势. 人民交通, 2020,(9): 84- 86. URL
	CHEN J. Analysis of the development trend of industrialization of urban intelligent traffic system in China. People's Transportation, 2020,(9): 84- 86. URL
2	王笑京. 中国智能交通发展回眸(二)——对中国智能交通起步有重要影响的几次国际交流. 中国交通信息化, 2019,(5): 15, 18- 25. doi: 10.13439/j.cnki.itsc.2019.05.001
	WANG X J. Review on the development of intelligent transportation in China (Ⅱ)—several international exchanges that have important influence on the start of intelligent transportation in China. China ITS Journal, 2019,(5): 15, 18- 25. doi: 10.13439/j.cnki.itsc.2019.05.001
3	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. Washington D. C., USA: IEEE Press, 2014: 580-587.
4	GIRSHICK R. Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2015: 1440-1448.
5	REN S, HE K, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137- 1149. doi: 10.1109/TPAMI.2016.2577031
6	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[EB/OL]. [2023-11-05]. https://arxiv.org/abs/1512.02325.
7	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]//Proceedings of the IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 2980-2988.
8	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 779-788.
9	WANG H W, YANG J D, GE F H, et al. Traffic sign detection algorithm based on improved YOLOv4[C]//Proceedings of the 6th IEEE Information Technology, Networking, Electronic and Automation Control Conference. Washington D. C., USA: IEEE Press, 2023: 142-147.
10	贾子豪, 王文青, 刘光灿. 改进YOLOv5的轻量化交通标志检测算法. 数据采集与处理, 2023, 38(6): 1434- 1444. doi: 10.16337/j.1004-9037.2023.06.017
	JIA Z H, WANG W Q, LIU G C. Improved lightweight traffic sign detection algorithm of YOLOv5. Journal of Data Acquisition and Processing, 2023, 38(6): 1434- 1444. doi: 10.16337/j.1004-9037.2023.06.017
11	WANG J F, CHEN Y, DONG Z K, et al. Improved YOLOv5 network for real-time multi-scale traffic sign detection. Neural Computing and Applications, 2023, 35(10): 7853- 7865. doi: 10.1007/s00521-022-08077-5
12	赵宏, 冯宇博. 基于CGS-Ghost YOLO的交通标志检测研究. 计算机工程, 2023, 49(12): 194- 204. doi: 10.19678/j.issn.1000-3428.0066520
	ZHAO H, FENG Y B. Research on traffic sign detection based on CGS-Ghost YOLO. Computer Engineering, 2023, 49(12): 194- 204. doi: 10.19678/j.issn.1000-3428.0066520
13	查超能, 罗素云, 何佳. 雨天下基于注意力机制与特征融合的交通标志识别. 重庆理工大学学报(自然科学版), 2023, 37(11): 20- 31. URL
	ZHA C N, LUO S Y, HE J. Traffic sign recognition based on attention mechanism and feature fusion in rainy weather. Journal of Chongqing University of Technology(Natural Science), 2023, 37(11): 20- 31. URL
14	谭鑫平, 高志辉, 韩航迪, 等. 基于改进YOLOv5的荧光图像细胞智能检测研究. 半导体光电, 2023, 44(5): 709- 716. doi: 10.16818/j.issn1001-5868.2023052801
	TAN X P, GAO Z H, HAN H D, et al. Intelligent detection of cells in fluorescence images based on improved YOLOv5. Semiconductor Optoelectronics, 2023, 44(5): 709- 716. doi: 10.16818/j.issn1001-5868.2023052801
15	HE K, ZHANG X, REN S, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1904- 1916. doi: 10.1109/TPAMI.2015.2389824
16	WANG C Y, BOCHKOVSKIY A, LIAO H Y M. Scaled-YOLOv4: scaling cross stage partial network[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2021: 13029-13038.
17	LIN T Y, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2017: 2117-2125.
18	LIU S, QI L, QIN H F, et al. Path aggregation network for instance segmentation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2018: 8759-8768.
19	HOWARD A G, ZHU M L, CHEN B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications[EB/OL]. [2023-11-05]. http://arxiv.org/abs/1704.04861v1.
20	MA N N, ZHANG X Y, ZHENG H T, et al. ShuffleNet V2: practical guidelines for efficient CNN architecture design[EB/OL]. [2023-11-05]. https://link.springer.com/chapter/10.1007/978-3-030-01264-9_8.
21	WANG P Q, CHEN P F, YUAN Y, et al. Understanding convolution for semantic segmentation[C]//Proceedings of the IEEE Winter Conference on Applications of Computer Vision. Washington D. C., USA: IEEE Press, 2018: 1451-1460.
22	WANG J Q, CHEN K, XU R, et al. CARAFE: content-aware ReAssembly of FEatures[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2019: 3007-3016.
23	WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[EB/OL]. [2023-11-05]. https://arxiv.org/abs/1807.06521.
24	ZHANG Q L, YANG Y B. SA-Net: shuffle attention for deep convolutional neural networks[C]//Proceedings of the ICASSP 2021 IEEE International Conference on Acoustics, Speech and Signal Processing. Washington D. C., USA: IEEE Press, 2021: 2235-2239.
25	LIU W, RABINOVICH A, BERG A C. ParseNet: looking wider to see better[EB/OL]. [2023-11-05]. http://arxiv.org/abs/1506.04579v2.
26	DAI Y M, GIESEKE F, OEHMCKE S, et al. Attentional feature fusion[C]//Proceedings of the IEEE Winter Conference on Applications of Computer Vision. Washington D. C., USA: IEEE Press, 2021: 3560-3569.
27	ZHANG J M, HUANG M T, JIN X K, et al. A real-time Chinese traffic sign detection algorithm based on modified YOLOv2. Algorithms, 2017, 10(4): 127. doi: 10.3390/a10040127
28	WANG Q L, WU B G, ZHU P F, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 11534-11542.
29	HU J, SHEN L, ALBANIE S, et al. Squeeze-and-excitation networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(8): 2011- 2023. doi: 10.1109/TPAMI.2019.2913372
30	吕禾丰, 陆华才. 基于YOLOv5算法的交通标志识别技术研究. 电子测量与仪器学报, 2021, 35(10): 137- 144. doi: 10.13382/j.jemi.B2104449
	LÜ H F, LU H C. Research on traffic sign recognition technology based on YOLOv5 algorithm. Journal of Electronic and Instrumentation, 2021, 35(10): 137- 144. doi: 10.13382/j.jemi.B2104449
31	WANG X J, TIAN Y Q, ZHENG K F, et al. C2Net-YOLOv5: a bidirectional Res2Net-based traffic sign detection algorithm. Computers, Materials & Continua, 2023, 77(2): 1949- 1965. doi: 10.32604/cmc.2023.042224

[1]	魏嵬, 丁香香, 郭梦星, 杨钊, 刘辉. 文本相似度计算方法综述[J]. 计算机工程, 2024, 50(9): 18-32.
[2]	朱凯, 李理, 张彤, 江晟, 别一鸣. 基于Transformer的多阶段运动模糊图像修复网络[J]. 计算机工程, 2024, 50(9): 276-285.
[3]	张天鹏, 韩晶, 吕学强. 基于多任务学习的超分辨率辅助小目标检测[J]. 计算机工程, 2024, 50(9): 304-312.
[4]	高煜宝, 文志诚. 基于注意力机制的双路解码器图像去噪方法[J]. 计算机工程, 2024, 50(9): 324-332.
[5]	张华青, 夏张涛, 陆晓庆, 童基均. 基于字形特征的血管外科命名实体识别[J]. 计算机工程, 2024, 50(8): 13-21.
[6]	张亚洲, 和玉, 戎璐, 王祥凯. 基于上下文知识增强型Transformer网络的抑郁检测[J]. 计算机工程, 2024, 50(8): 75-85.
[7]	高伟, 李帅龙, 茆琳, 王磊, 李颖颖, 韩林. 一种基于TVM的算子生成加速策略[J]. 计算机工程, 2024, 50(8): 353-362.
[8]	王宇, 祁琦, 王纯, 许才. 储能变流器信号高精度故障诊断方法[J]. 计算机工程, 2024, 50(8): 389-396.
[9]	牛瑞婷, 严天峰, 高锐, 王映植. 低信噪比下基于深度学习TCNN-MobileNet的调制识别[J]. 计算机工程, 2024, 50(7): 204-215.
[10]	肖慈, 徐杨, 张永丹, 冯明文, 黄易仟. 结合注意力和低光增强的夜间语义分割[J]. 计算机工程, 2024, 50(7): 271-281.
[11]	张诗婧, 莫绪涛, 赵行, 董杨林. 基于球面折反射成像和YOLOv7的内螺纹缺陷检测[J]. 计算机工程, 2024, 50(7): 282-292.
[12]	贵向泉, 刘世清, 李立, 秦庆松, 李唐艳. 基于改进YOLOv8的景区行人检测算法[J]. 计算机工程, 2024, 50(7): 342-351.
[13]	徐明亮, 李芳媛, 马浩然, 何飞. 大规模神经记录的峰电位聚类算法(特邀)[J]. 计算机工程, 2024, 50(6): 1-34.
[14]	魏琢艺, 罗迈, 李文兵, 曾远松, 余伟江, 杨跃东. 基于多源域适应的单细胞智能分类[J]. 计算机工程, 2024, 50(6): 48-55.
[15]	李子杰, 周菊香, 韩晓瑜, 甘健侯, 鹿泽光, 王俊. 序列特征与学习过程融合的知识追踪模型[J]. 计算机工程, 2024, 50(6): 77-85.

选择文件类型/文献管理软件名称

选择包含的内容