基于改进YOLOv8的实时坑槽检测算法

doi:10.19678/j.issn.1000-3428.0069671

摘要/Abstract

摘要：

针对道路坑槽检测中存在坑槽大小不同、形状不规则导致的特征提取不完全及图像拍摄不满足道路检测车的视角问题, 收集并制作不同来源、视角和像素分辨率的坑槽数据集, 并对模型进行改进。首先在Backbone部分的C2f结构中引入DCNv3, 以获取更丰富完整的坑槽特征; 其次融合压缩和激励(SE)模块的注意力机制, 以提高对坑槽特征的提取能力; 然后在Neck部分融合双向特征金字塔网络(BiFPN)结构, 降低网络的计算量; 最后使用Focal-EIoU作为改进模型的损失函数, 降低复杂背景对网络检测性能的影响。改进后的YOLOv8-master网络相较于未改进前的网络, 坑槽检测精度提高了4.06%, 检测速度提高了85帧/s, 浮点运算量降低了19.54%。结果表明, 所提出的改进方法能有效提高原网络检测坑槽的性能, 相比目前主流的目标检测算法, 具有一定的先进性。

关键词: 坑槽检测, 可变形卷积, 压缩和激励模块, 双向特征金字塔网络, Focal-EIoU损失函数

Abstract:

To address the issues of incomplete feature extraction due to varying sizes and irregular shapes of potholes and the problem of image capturing not satisfying the perspective of road inspection vehicles during pothole detection, we have collected and created a pothole dataset from diverse sources, perspectives, and pixel resolutions, and further improved the model. First, we introduced DCNv3 into the C2f structure of the Backbone section to capture richer and more complete pothole features. Second, we integrated the attention mechanism of Squeeze-and-Excitation (SE) model to enhance the ability to extract pothole features. Third, we fused the BiFPN structure in the Neck section to reduce the computational complexity of the network. Finally, we used Focal-EIoU as the loss function of the improved model to minimize the impact of complex backgrounds on network detection performance. Compared to the unimproved network, the enhanced YOLOv8-master network achieved a 4.06% improvement in pothole detection accuracy, a detection speed boost to 85 frame per second, and a 19.54% reduction in floating-point operations. The results demonstrate that the proposed improvement method effectively enhances the original network's performance in detecting potholes and possesses certain advancements compared to currently mainstream object detection algorithms.

Key words: pothole detection, deformable convolution, Squeeze-and-Excitation (SE) module, Bidirectional Feature Pyramid Network (BiFPN), Focal-EIoU loss function

马荣贵, 黄训燕, 董世浩. 基于改进YOLOv8的实时坑槽检测算法[J]. 计算机工程, 2025, 51(11): 226-234.

MA Ronggui, HUANG Xunyan, DONG Shihao. Real-time Pothole Detection Algorithm Based on Improved YOLOv8[J]. Computer Engineering, 2025, 51(11): 226-234.

https://www.ecice06.com/CN/Y2025/V51/I11/226

图/表 18

图1 YOLOv8网络结构

Fig.1 YOLOv8 network structure

图2 C2f模块结构图

Fig.2 C2f module structure diagram

图3 改进后得到的YOLOv8-master网络结构图

Fig.3 Improved YOLOv8-master network structure diagram

图4 常规卷积与可变形卷积的采样点

Fig.4 Sampling points of conventional convolution and deformable convolutions

图5 可变形卷积计算过程

Fig.5 Calculation process of deformable convolution

图6 C2f融合可变形卷积结构

Fig.6 C2f fusion deformable convolutional structure

图7 SE网络结构图

Fig.7 SE network structure diagram

图8 注意力机制的添加位置

Fig.8 Location of added attention mechanism

图9 改进的BiFPN结构

Fig.9 Improved BiFPN structure

图10 半自动标注流程

Fig.10 Semi-automatic annotation process

图11 坑槽目标与背景差异不大状态下的检测结果对比

Fig.11 Comparison of detection results for states with little difference between pit targets and background

图12 存在多目标、尺寸差异较大状态下的检测结果对比

Fig.12 Comparison of detection results for states with multiple targets and significant size differences

图13 坑槽内部有积水状态下的检测结果对比

Fig.13 Comparison of detection results for water accumulation inside pit and groove

图14 坑槽目标部分有积水、部分没有积水状态下的检测结果对比

Fig.14 Comparison of detection results for water accumulation in target part and no water accumulation in some parts of pit

图15 阳光阴影、多个目标尺寸差异大状态下的检测结果对比

Fig.15 Comparison of detection results for multiple targets with large size differences in sunshine shadows

参考文献 30

1	王艾迪, 彭一川, 郎洪, 等. 基于YOLOX-Transformer两步模型的路面坑槽提取方法. 中国公路学报, 2023, 36 (12): 304- 317.
	WANG A D , PENG Y C , LANG H , et al. Pavement pothole extraction based on YOLOX-Transformer two-step model. China Journal of Highway and Transport, 2023, 36 (12): 304- 317.
2	沙爱民, 童峥, 高杰. 基于卷积神经网络的路表病害识别与测量. 中国公路学报, 2018, 31 (1): 1- 10.
	SHA A M , TONG Z , GAO J . Recognition and measurement of pavement disasters based on convolutional neural networks. China Journal of Highway and Transport, 2018, 31 (1): 1- 10.
3	王大为, 吕浩天, 汤伏蛟, 等. 三维探地雷达道路隐性病害检测分析与数字化技术综述. 中国公路学报, 2023, 36 (3): 1- 19.
	WANG D W , LÜ H T , TANG F J , et al. Road structural defects detection and digitalization based on 3D ground penetrating radar technology: a state-of-the-art review. China Journal of Highway and Transport, 2023, 36 (3): 1- 19.
4	NIENABER S, BOOYSEN M J, KROON R. Detecting potholes using simple image processing techniques and real-world footage[C]//Proceedings of 34th Annual Southern African Transport Conference (SATC 2015). Singapore, Singapore: Springer, 2021: 153-164.
5	JANG D W , PARK R H . Pothole detection using spatio-temporal saliency. IET Intelligent Transport Systems, 2016, 10 (9): 605- 612. doi: 10.1049/iet-its.2016.0006
6	AKAGIC A, BUZA E, OMANOVIC S. Pothole detection: an efficient vision based method using RGB color space image segmentation[C]//Proceedings of the 40th International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO). Washington D. C., USA: IEEE Press, 2017: 1104-1109.
7	姜百浩, 刘静, 仇大伟, 等. 深度学习在脊柱图像分割中的应用综述. 计算机工程, 2024, 50 (3): 1- 15. doi: 10.19678/j.issn.1000-3428.0067502
	JIANG B H , LIU J , QIU D W , et al. Review of deep learning applications in spinal image segmentation. Computer Engineering, 2024, 50 (3): 1- 15. doi: 10.19678/j.issn.1000-3428.0067502
8	贵向泉, 刘世清, 李立, 等. 基于改进YOLOv8的景区行人检测算法. 计算机工程, 2024, 50 (7): 342- 351. doi: 10.19678/j.issn.1000-3428.0068125
	GUI X Q , LIU S Q , LI L , et al. Pedestrian detection algorithm for scenic spots based on improved YOLOv8. Computer Engineering, 2024, 50 (7): 342- 351. doi: 10.19678/j.issn.1000-3428.0068125
9	黄开基, 杨华. 基于深度学习特征的二维图像匹配算法综述. 计算机工程, 2024, 50 (10): 16- 34. doi: 10.19678/j.issn.1000-3428.0068580
	HUANG K J , YANG H . Review of 2D image matching algorithms based on deep learning features. Computer Engineering, 2024, 50 (10): 16- 34. doi: 10.19678/j.issn.1000-3428.0068580
10	连哲, 殷雁君, 云飞, 等. 基于深度学习的自然场景文本检测综述. 计算机工程, 2024, 50 (3): 16- 27. doi: 10.19678/j.issn.1000-3428.0067427
	LIAN Z , YLN Y J , YUN F , et al. Review of natural scene text detection based on deep learning. Computer Engineering, 2024, 50 (3): 16- 27. doi: 10.19678/j.issn.1000-3428.0067427
11	DHIMAN A , KLETTE R . Pothole detection using computer vision and learning. IEEE Transactions on Intelligent Transportation Systems, 2020, 21 (8): 3536- 3550. doi: 10.1109/TITS.2019.2931297
12	DHIMAN A, CHIEN H J, KLETTE R. A multi-frame stereo vision-based road profiling technique for distress analysis[C]//Proceedings of the 15th International Symposium on Pervasive Systems, Algorithms and Networks (I-SPAN). Washington D. C., USA: IEEE Press, 2018: 7-14.
13	UKHWAH E N, YUNIARNO E M, SUPRAPTO Y K. Asphalt pavement pothole detection using deep learning method based on YOLO neural network[C]//Proceedings of the International Seminar on Intelligent Technology and Its Applications (ISITIA). Washington D. C., USA: IEEE Press, 2019: 35-40.
14	PAN Y F , ZHANG X F , CERVONE G , et al. Detection of asphalt pavement potholes and cracks based on the unmanned aerial vehicle multispectral imagery. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018, 11 (10): 3701- 3712. doi: 10.1109/JSTARS.2018.2865528
15	LI C, LI L, GENG Y, et al. YOLOv6 v3. 0: a full-scale reloading[EB/OL]. (2023-01-13)[2024-02-29]. https://arxiv.org/abs/2301.05586.
16	DAI J F, QI H Z, XIONG Y W, et al. Deformable convolutional networks[C]//Proceedings of the IEEE International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2017: 764-773.
17	HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2018: 7132-7141.
18	ZHU X Z, HU H, LIN S, et al. Deformable ConvNets V2: more deformable, better results[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2019: 9308-9316.
19	汤志康, 武毓琦, 李春英, 等. 基于知识图谱卷积网络的学习资源推荐. 计算机工程, 2024, 50 (9): 153- 160. doi: 10.19678/j.issn.1000-3428.0068409
	TANG Z K , WU Y Q , LI C Y , et al. Recommendation of learning resource based on knowledge graph convolutional network. Computer Engineering, 2024, 50 (9): 153- 160. doi: 10.19678/j.issn.1000-3428.0068409
20	邓中港, 代刚, 吴湘宁, 等. 基于注意力机制和可变形卷积的金属表面细微不规则损伤的图像识别模型. 计算机工程与科学, 2023, 45 (1): 127- 135.
	DENG Z G , DAI G , WU X N , et al. An image recognition model for minor and irregular damage on metal surface based on attention mechanism and deformable convolution. Computer Engineering[WT《Times New Roman》] & Science, 2023, 45 (1): 127- 135.
21	蔡俊民, 梁正友, 孙宇, 等. 基于可变形三维图卷积的轻量级点云分类研究. 计算机工程, 2024, 50 (9): 255- 265. doi: 10.19678/j.issn.1000-3428.0067589
	CAI J M , LIANG Z Y , SUN Y , et al. Research on lightweight point cloud classification based on deformable 3D graph convolution. Computer Engineering, 2024, 50 (9): 255- 265. doi: 10.19678/j.issn.1000-3428.0067589
22	周新, 郭敬楠, 宁博, 等. IntSE: 特征增强的知识图谱补全方法. 小型微型计算机系统, 2023, 44 (9): 1961- 1965.
	ZHOU X , GUO J N , NING B . IntSE: feature enhanced knowledge graph completion method. Journal of Chinese Computer Systems, 2023, 44 (9): 1961- 1965.
23	刘强, 张道畅. 结合SE的密集卷积生成对抗网络图像修复方法. 小型微型计算机系统, 2022, 43 (5): 1056- 1060.
	LIU Q , ZHANG D C . Dense convolution generative adversarial network image inpainting method with SENet. Journal of Chinese Computer Systems, 2022, 43 (5): 1056- 1060.
24	赵璐璐, 王学营, 张翼, 等. 基于YOLOv5s融合SENet的车辆目标检测技术研究. 图学学报, 2022, 43 (5): 776- 782.
	ZHAO L L , WANG X Y , ZHANG Y , et al. Vehicle target detection based on YOLOv5s fusion SENet. Journal of Graphics, 2022, 43 (5): 776- 782.
25	LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2017: 936-944.
26	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.
27	TAN M X, PANG R M, LE Q V. EfficientDet: scalable and efficient object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2020: 10781-10790.
28	彭道刚, 潘俊臻, 王丹豪, 等. 基于改进YOLOv5的电厂管道油液泄漏检测. 电子测量与仪器学报, 2022, 36 (12): 200- 209.
	PENG D G , PAN J Z , WANG D H , et al. Oil leakage detection of pipelines of power plants based on improved YOLOv5. Journal of Electronic Measurement and Instrumentation, 2022, 36 (12): 200- 209.
29	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 (ICCV). Washington D. C., USA: IEEE Press, 2017: 2999-3007.
30	ZHANG Y F , REN W Q , ZHANG Z , et al. Focal and efficient IOU loss for accurate bounding box regression. Neurocomputing, 2022, 506, 146- 157. doi: 10.1016/j.neucom.2022.07.042

[1]	彭菊红, 张弛, 高谦, 张光明, 谈栋华, 赵明俊. 基于改进的YOLOv8算法的钢材缺陷检测[J]. 计算机工程, 2025, 51(7): 152-160.
[2]	张天旭, 黄慧, 黄丙仓, 马燕, 徐傲, 李晓艳, 周孝雯, 刘之之. 基于多尺度聚合与高分辨率增强的CTA脑血管分割模型[J]. 计算机工程, 2025, 51(4): 37-46.
[3]	赵南南, 高翡晨. 基于改进YOLOv8的交通场景实例分割算法[J]. 计算机工程, 2025, 51(1): 198-207.
[4]	徐芳芯, 樊嵘, 马小陆. 面向拥挤行人检测的改进YOLOv7算法[J]. 计算机工程, 2024, 50(3): 250-258.
[5]	候瑞环, 杨喜旺, 王智超, 高佳鑫. 一种基于YOLOv4-TIA的林业害虫实时检测方法[J]. 计算机工程, 2022, 48(4): 255-261.
[6]	秦鹏, 唐川明, 刘云峰, 张建林, 徐智勇. 基于改进YOLOv3的红外目标检测方法[J]. 计算机工程, 2022, 48(3): 211-219.
[7]	尚佳童, 雷涛, 张栋, 杜晓刚, 翟钰杰. 面向刻蚀图像分割的轻量可变形编解码网络[J]. 计算机工程, 2022, 48(12): 203-211,217.
[8]	刘振国, 李钊, 宋滕滕, 何益智. 结合可变形卷积与双边网格的立体匹配网络[J]. 计算机工程, 2022, 48(12): 241-247,254.
[9]	李富豪, 赵希梅. 基于D-Unet神经网络的鼻腔鼻窦肿瘤分割算法[J]. 计算机工程, 2022, 48(1): 281-287.
[10]	朱灵灵, 高超, 陈福才. 基于轻量级卷积神经网络的人脸检测算法[J]. 计算机工程, 2021, 47(7): 273-280.
[11]	包俊, 刘宏哲. 融合可变形卷积网络的鱼眼图像目标检测[J]. 计算机工程, 2021, 47(4): 248-255.
[12]	喻清挺, 喻维超, 喻国平. 基于改进R-FCN的交通标志检测[J]. 计算机工程, 2021, 47(12): 285-290,298.
[13]	黄凤琪, 陈明, 冯国富. 基于可变形卷积的改进YOLO目标检测算法[J]. 计算机工程, 2021, 47(10): 269-275,282.
[14]	翟强, 王陆洋, 殷保群, 彭思凡, 邢思思. 基于尺度自适应卷积神经网络的人群计数算法[J]. 计算机工程, 2020, 46(2): 250-254,261.

选择文件类型/文献管理软件名称

选择包含的内容