改进YOLOv5的复杂环境道路目标检测方法

doi:10.3778/j.issn.1002-8331.2304-0251

摘要/Abstract

摘要： 为解决复杂环境下道路目标检测任务中由于目标尺度变化多样、密集遮挡以及光照不均匀等导致的漏检问题，提出了一种基于YOLOv5的道路目标检测改进方法CTC-YOLO（contextual transformer and convolutional block attention module based on YOLOv5）。针对小目标，改进网络检测头结构，增加多尺度目标检测层，提高小目标检测精度。为了充分利用输入的上下文信息，在特征提取部分引入上下文变换模块（contextual transformer networks，CoTNet），设计了CoT3模块，引导动态注意力矩阵学习，提高视觉表征能力。在Neck部分的C3模块集成卷积块注意力模型（convolutional block attention module，CBAM），以在各种复杂的场景中找到注意力区域。为进一步验证CTC-YOLO方法，采取了一些有用的策略，如模型集成位置选择和对比其他注意力机制。实验结果表明，在公开数据集KITTI、Cityscapes以及BDD100K上mAP@0.5分别达到89.6%、46.1%和57.0%，相较基线模型分别提高3.1个百分点、2.0个百分点和1.2个百分点。与其他模型相比，检测效率更高，有效改善了复杂环境中的目标检测问题。

关键词: 复杂环境, 目标检测, YOLOv5, 注意力机制

Abstract: To solve the problem of missed detection in road object detection tasks in complex environments due to diverse target scale changes, dense occlusion and uneven lighting, an improved method for road object detection CTC-YOLO（context transformer and convolutional block attention module based on YOLOv5） is proposed. Firstly, for small targets, improve the network detection head structure, add a multi-scale target detection layer, and improve the accuracy of small target detection. Secondly, in order to fully utilize the input contextual information, introduce a context transformer networks（CoTNet） module in the feature extraction section, and design a CoT3 module to guide dynamic attention matrix learning and improve visual representation ability. Finally, the C3 module in the Neck section integrates the convolutional block attention module（CBAM） to locate attention regions in complex scenes. To further validate the CTC-YOLO method proposed in this paper, some useful strategies are adopted, such as model integration position selection and comparison with other attention mechanisms. The experimental results show that the mAP@0.5 on the publicly available datasets KITTI, Cityscapes and BDD100K reaches 89.6%, 46.1% and 57.0%, respectively, which are 3.1, 2.0 and 1.2 percentage points higher than the baseline model, respectively. Compared with other models, the detection efficiency is higher and effectively improves the problem of object detection in complex environments.

Key words: complex environment, target detection, YOLOv5, attentional mechanism

袁磊, 唐海, 陈彦蓉, 高刃, 吴文欢. 改进YOLOv5的复杂环境道路目标检测方法[J]. 计算机工程与应用, 2023, 59(16): 212-222.

YUAN Lei, TANG Hai, CHEN Yanrong, GAO Ren, WU Wenhuan. Improved YOLOv5 for Road Target Detection in Complex Environments[J]. Computer Engineering and Applications, 2023, 59(16): 212-222.

参考文献

[1] 中华人民共和国公安部.全国机动车保有量达4.17亿辆，驾驶人超过5亿人[EB/OL].（2023-01-11）[2023-04-19].http：//app.mps.gov.cn/gdnps/pc/content.jsp?id=8837602.
The Ministry of Public Security of the People’s Republic of China.The number of motor vehicles in China reached 417 million，with more than 500 million drivers[EB/OL].（2023-01-11）[2023-04-19].http：//app.mps.gov.cn/gdnps/pc/content.jsp?id=8837602.
[2] LIU W，ANGUELOV D，ERHAN D，et al.SSD：single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision，Amsterdam，Oct 11-14，2016.Cham：Springer，2016：21-37.
[3] REDMON J，DIVVALA S，GIRSHICK R，et al.You only look once：unified，real-time object detection[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition，2016：779-788.
[4] REDMON J，FARHADI A.YOLO9000：better，faster，stronger[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition，2017：7263-7271.
[5] REDMON J，FARHADI A.YOLOv3：an incremental improvement[J].arXiv：1804.02767，2018.
[6] BOCHKOVSKIY A，WANG C Y，LIAO H Y M.YOLOv4：optimal speed and accuracy of object detection[J].arXiv：2004.10934，2020.
[7] LI C，LI L，JIANG H，et al.YOLOv6：a single-stage object detection framework for industrial applications[J].arXiv：2209.02976，2022.
[8] WANG C Y，BOCHKOVSKIY A，LIAO H Y M.YOLOv7：trainable bag-of-free-bies sets new state-of-the-art for real-time object detectors[J].arXiv：2207.02696，2022.
[9] GE Z，LIU S，WANG F，et al.YOLOx：exceeding YOLO series in 2021[J].arXiv：2107.08430，2021.
[10] GIRSHICK R，DONAHUE J，DARRELL T，et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition，2014：580-587.
[11] GIRSHICK R.Fast R-CNN[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision，2015：1440-1448.
[12] REN S，HE K，GIRSHICK R，et al.Faster R-CNN：towards real-time object detection with region proposal networks[C]//Advances in Neural Information Processing Systems 28：Annual Conference on Neural Information Processing Systems，2015：91-99.
[13] HE K，GKIOXARI G，DOLLáR P，et al.Mask R-CNN[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision，2017：2961-2969.
[14] 韦强，胡晓阳，赵虹鑫.改进YOLOv5的交通标志检测方法[J].计算机工程与应用，2023，59（13）：229-237.
WEI Q，HU X Y，ZHAO H X.Improving traffic sign detection method for YOLOv5[J].Computer Engineering and Applications，2023，59（13）：229-237.
[15] 朱瑞鑫，杨福兴.运动场景下改进YOLOv5小目标检测算法[J].计算机工程与应用，2023，59（10）：196-203.
ZHU R X，YANG F X.Improved YOLOv5 small object detection algorithm in moving scenes[J].Computer Engineering and Applications，2023，59（10）：196-203.
[16] ZHANG Y，GUO Z，WU J，et al.Real-time vehicle detection based on improved YOLOv5[J].Sustainability，2022，14（19）：12274.
[17] LI Z.Road aerial object detection based on improved YOLOv5[J].Journal of Physics：Conference Series，2022，2171（1）：012039.
[18] JIN Z Z，ZHENG Y F.Research on application of improved YOLO V3 algorithm in road target detection[J].Journal of Physics：Conference Series，2020，1654（1）：012060.
[19] 江德港，江智，黄子杰，等.基于Efficientnet的无人机车辆目标检测算法[J].计算机工程与应用，2023，59（12）：228-234.
JIANG D G，JIANG Z，HUANG Z J，et al.UAN vehicle object detection algorithm based on Efficientnet[J].Computer Engineering and Applications，2023，59（12）：228-234.
[20] 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 Y X，WU X R，XUE G G.Multi-target real-time detection for road traffic signs based on deep learning[J].Journal of Guangxi Normal University（Natural Science Edition），2020，38（2）：96-106.
[21] LI Y，YAO T，PAN Y，et al.Contextual transformer networks for visual recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2023，45（2）：1489-1500.
[22] WOO S，PARK J，LEE J Y，et al.CBAM：convolutional block attention module[C]//Proceedings of the 15th European Conference on Computer Vision，2018：3-19.
[23] GEIGER A，LENZ P，URTASUN R.Are we ready for autonomous driving? The KITTI vision benchmark suite[C]//Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition，2012：3354-3361.
[24] CORDTS M，OMRAN M，RAMOS S，et al.The Cityscapes dataset for semantic urban scene understanding[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition，2016：3213-3223.
[25] YU F，CHEN H，WANG X，et al.BDD100K：a diverse driving dataset for heterogeneous multitask learning[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2020：2636-2645.
[26] LIU S，QI L，QIN H，et al.Path aggregation network for instance segmentation[C]//Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition，2018：8759-8768.
[27] ZHU X，LYU S，WANG X，et al.TPH-YOLOv5：improved YOLOv5 based on transformer prediction head for object detection on drone-captured scenarios[C]//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision，2021：2778-2788.
[28] VASWANI A，SHAZEER N，PARMAR N，et al.Attention is all you need[C]//Advances in Neural Information Processing Systems 30：Annual Conference on Neural Information Processing Systems，2017：5998-6008.
[29] HU J，SHEN L，SUN G.Squeeze-and-excitation networks[C]//Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition，2018：7132-7141.
[30] 易见兵，黄苏泉，曹锋，等.基于改进的YOLOX血细胞检测算法研究[J].电子测量技术，2022，45（22）：177-184.
YI J B，HUANG S Q，CAO F，et al.Research on improved blood cell detection algorithm based on YOLOX[J].Electronic Measurement Technology，2022，45（22）：177-184.
[31] GRIFFIN G，HOLUB A，PERONA P.Caltech-256 object category dataset[J].California Institute of Technology，2007.
[32] ZHENG Z，WANG P，LIU W，et al.Distance-IoU loss：faster and better learning for bounding box regression[C]//Proceedings of the 34th AAAI Conference on Artificial Intelligence，2020：12993-13000.
[33] LEE Y，PARK J.CenterMask：real-time anchor-free instance segmentation[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2020：13906-13915.
[34] LIU Y，SHAO Z，HOFFMANN N.Global attention mechanism：retain information to enhance channel-spatial interactions[J].arXiv：2112.05561，2021.
[35] RAO Y，ZHAO W，TANG Y，et al.HorNet：efficient high-order spatial interactions with recursive gated convolutions[C]//Advances in Neural Information Processing Systems 35，2022：10353-10366.
[36] LIU Z，MAO H，WU C Y，et al.A ConvNet for the 2020s[C]//Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2022：11976-11986.
[37] LIU Z，LIN Y，CAO Y，et al.Swin transformer：hierarchical vision transformer using shifted windows[C]//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision，2021：10012-10022.