基于YOLOv5的轻量化桥梁缺陷识别

doi:10.13229/j.cnki.jdxbgxb.20250539

Abstract

Abstract:

Based on the YOLOv5 algorithm， the feature extraction process was optimized by introducing the FasterNet lightweight network structure to reduce the computational complexity of the model. Combining Convolutional Block Attention Module （CBAM） attention mechanism to improve the model's attention to defect features. Adopting the SIoU loss function to improve the accuracy of bounding box regression and enhance localization accuracy. The experimental results show that the improved model has achieved significant results on the self built bridge defect dataset， with accuracy and mAP reaching 81.2% and 71.5%， respectively， which is a significant improvement compared to the benchmark model. This study provides technical support for the intelligent detection of bridge defects， which is of great significance for promoting the digital transformation of bridge maintenance management.

Key words: engineering of communications and transportation system, defect detection, deep learning, UAV perspective

CLC Number:

U447

Lin-hong WANG,Yu-yang LIU,Zi-yu LIU,Ying-jia LU,Yu-heng ZHANG,Gui-shu HUANG. Defect recognition of lightweight bridges based on YOLOv5[J].Journal of Jilin University(Engineering and Technology Edition), 2025, 55(9): 2958-2968.

Figures/Tables 17

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Fig.7

Fig.8

Fig.9

Fig.10

Fig.11

Fig.12

Fig.13

Table 1

Table 2

Table 3

Table 4

References 26

[1]	丁威, 夏哲, 舒江鹏, 等. 基于负压吸附爬壁机器人和Transformer的混凝土桥塔裂缝识别检测[J]. 中国公路学报, 2024, 37(2): 53-64.
	Ding Wei, Xia Zhe, Shu Jiang-peng, et al. Recognition and detection of concrete bridge tower cracks using a negative pressure adhesion wall-climbing robot and Transformer[J]. China Journal of Highway and Transport, 2024, 37(2): 53-64.
[2]	刘凯, 李家坡, 邓智允, 等. 无人机桥梁检测技术的经济效益分析[J]. 南昌工程学院学报, 2018(3): 51-53, 80.
	Liu Kai, Li Jia-po, Deng Zhi-yun, et al. Benefit analysis of unmanned aerial vehicle bridge detection technology[J]. Journal of Nanchang Institute of Technology, 2018(3): 51-53, 80.
[3]	邹金欢. 无人机在输电线路巡检中的应用及效益分析[J]. 中国战略新兴产业, 2023(36): 153-155.
	Zou Jin-huan. Application and benefit analysis of drones in transmission line inspection[J]. China Strategic Emerging Industry, 2023(36): 153-155.
[4]	Oppenheim D, Shani G, Erlich O, et al. Using deep learning for image-based potato tuber disease detection[J]. Phytopathology, 2019, 109(6): 1083-1087.
[5]	Wang J X, Su Y H, Yao J H, et al. Apple rapid recognition and processing method based on an improved version of YOLOv5[J]. Ecological Informatics, 2023, 77: No.102196.
[6]	贾雪莹, 赵春江, 周娟, 等. 基于改进YOLOv7模型的柑橘表面缺陷在线检测[J]. 农业工程学报, 2023, 39(23): 142-151.
	Jia Xue-ying, Zhao Chun-jiang, Zhou Juan, et al. Online detection of citrus surface defects using improved YOLOv7 modeling[J]. Transactions of the Chinese Society of Agricultural Engineering, 2023, 39(23): 142-151.
[7]	王庞伟. 基于金字塔特征融合的二阶段三维点云车辆检测[J]. 交通运输系统工程与信息, 2022, 22(5):107-116.
	Wang Pang-wei. Pyramid-feature-fusion-based two-stage vehicle detection via 3D point cloud[J]. Journal of Transportation Systems Engineering and Information Technology, 2022, 22(5): 107-116.
[8]	张河山, 范梦伟, 谭鑫, 等. 基于改进YOLOX的无人机航拍图像密集小目标车辆检测[J]. 吉林大学学报: 工学版, 2025, 55(4): 1307-1318.
	Zhang He-shan, Fan Meng-wei, Tan Xin, et al. Dense small target vehicle detection in UAV aerial images based on improved YOLOX[J]. Journal of Jilin University (Engineering and Technology Edition), 2025, 55(4): 1307-1318.
[9]	Yang Li-na, Chen Gang, Wen-yan Ci. Multiclass objects detection algorithm using DarkNet-53 and DenseNet for intelligent vehicles[J] EURASIP Journal on Advances in Signal Processing, 2023, 2023(1): 1-18.
[10]	Cao F, Xing B, Luo J C, et al. An efficient object detection algorithm based on improved YOLOv5 for High-Spatial-Resolution remote sensing images[J]. Remote Sensing, 2023, 15(15):No. 3755.
[11]	许晓阳, 高重阳. 改进YOLOv7-tiny的轻量级红外车辆目标检测算法[J]. 计算机工程与应用, 2024, 60(1): 74-83.
	Xu Xiao-yang, Gao Chong-yang. Improved YOLOv7-tiny lightweight infrared vehicle target detection algorithm[J]. Computer Engineering and Applications, 2024, 60(1):74-83.
[12]	彭卫兵, 张明见, 全柳萌, 等. 基于改进Mask R-CNN的钢桥多缺陷智能识别[J]. 中国公路学报, 2024, 37(2): 100-109.
	Peng Wei-bing, Zhang Ming-jian, Quan Liu-meng, et al. Intelligent recognition of multiple diseases in steel bridges based on improved Mask R-CNN[J]. China Journal of Highway and Transport, 2024, 37(2): 100-109.
[13]	陈昭俊. 二阶段锚框和类均衡损失的遥感图像目标检测[J]. 图学学报, 2023,44(2): 249-259.
	Chen Zhao-jun. Object detection in remote sensing image based on two-stage anchor and class balanced loss[J]. Journal of Graphics, 2023,44(2): 249-259.
[14]	蒋燕芳. 基于图像处理与深度学习的RC桥梁表观病害识别[D]. 重庆: 重庆大学土木工程学院, 2019.
	Jiang Yan-fang. Identification of apparent defects in RC bridge with image processing and deep learning[D]. Chongqing:College of Civil Engineering, Chongqing University, 2019.
[15]	Zeng G, Wu Z Z, Xu L P, et al. Efficient vision Transformer YOLOv5 for accurate and fast traffic fign detection[J]. Electronics, 2024, 13(5):No. 880.
[16]	徐鑫. 基于YOLOv5的医学图像病变区域检测研究与应用[D]. 镇江: 江苏大学计算机科学与通信工程学院, 2023.
	Xu Xin. Research and application of medical image lesion region detection based on YOLOv5[D]. Zhenjiang:School of Computer Science and Communication Engineering, Jiangsu University, 2023.
[17]	钟源, 袁家政, 李鸿天, 等. 改进YOLOv5的智慧课堂人脸检测算法[J] .计算机工程与应用, 2024, 60(11): 251-257.
	Zhong Yuan, Yuan Jia-zheng, Li Hong-tian, et al. Intelligent classroom face detection algorithm with improved YOLOv5[J]. Computer Engineering and Applications, 2024, 60(11): 251-257.
[18]	Rejin V, Sambath M. YOLOv8: a novel object detection algorithm with enhanced performance and robustness[C]∥International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS), Chennai, India, 2024: 1-6.
[19]	Chen J R, Kao S H, He H, et al. Run, don't walk: chasing higher FLOPS for faster neural networks [C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR),Vancouver, Canada, 2023: 12021-12031.
[20]	Chen J, He T, Zhuo W, et al. TVConv: efficient translation variant convolution for layout-aware visual processing [C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, USA, 2022: 12538-12548.
[21]	张楠楠, 张晓, 白铁成, 等. 基于CBAM-YOLOv7的自然环境下棉叶病虫害识别方法[J]. 农业机械学报, 2023, 54(201): 239-244.
	Zhang Nan-nan, Zhang Xiao, Bai Tie-cheng, et al. Pest and disease identification method for cotton leaves in natural environments based on CBAM-YOLOv7[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(201): 239-244.
[22]	Woo S, Park J, Lee J Y, et al. CBAM: convolutional block attention module[J]. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2018, 11211: 3-19.
[23]	Hamid R, Nathan T, Junyoung G, et al. Generalized intersection over union: a metric and a loss for bounding box regression[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 658-666.
[24]	Gevorgyan Z. SIoU Loss: more powerful learning for bounding box regression[DB/OL]. [2025-06-05]. .
[25]	王一旭, 肖小玲, 王鹏飞, 等. 改进YOLOv5s的小目标烟雾火焰检测算法[J]. 计算机工程与应用, 2023, 59(1): 72-81.
	Wang Yi-xu, Xiao Xiao-ling, Wang Peng-fei, et al. Improved YOLOv5s small target smoke and fire detection algorithm[J]. Computer Engineering and Applications, 2023, 59(1): 72-81.
[26]	孙长亮, 何峻, 肖怀铁. 基于ROC曲线的目标识别性能评估方法[J]. 雷达科学与技术, 2007, 5(1): 17-21.
	Sun Chang-liang, He Jun, Xiao Huai-tie. A new performance evaluation method based on ROC curve[J]. Radar Science and Technology, 2007, 5(1): 17-21.

Related Articles 15

[1]	Jing LIAN,Ji-bao ZHANG,Ji-zhao LIU,Jia-jun ZHANG,Zi-long DONG. Text-based guided face image inpainting [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(8): 2732-2740.
[2]	Yuan-ning LIU,Xing-zhe WANG,Zi-yu HUANG,Jia-chen ZHANG,Zhen LIU. Stomach cancer survival prediction model based on multimodal data fusion [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(8): 2693-2702.
[3]	Jiang-bo YU,Jian-cheng WENG,Peng-fei LIN,Yu-xing SUN,Jiao-long CHAI. Passenger flow prediction model of external transportation hub based on hybrid Transformer [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(7): 2251-2259.
[4]	Shu-shan CHAI,Zhi-qiang ZHOU,Hai-tao LI,Jiong-yang XU. Real-time road network traffic anomaly incident detection based on graph spatial-temporal pattern learning network [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(7): 2145-2161.
[5]	Jing-shu YUAN,Wu LI,Xing-yu ZHAO,Man YUAN. Semantic matching model based on BERTGAT-Contrastive [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(7): 2383-2392.
[6]	Hui-zhi XU,Dong-sheng HAO,Xiao-ting XU,Shi-sen JIANG. Expressway small object detection algorithm based on deep learning [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(6): 2003-2014.
[7]	Hong-zhuan ZHAO,Ze-jian WU,Xin ZHANG,Sheng-wen SHI,Wen-yong LI,Xin ZHAN,En-yong XU,Jia-ming WANG. Curve lattice model for connected commercial vehicles based on density dispersion and information transmission delay [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(6): 2015-2029.
[8]	Ru-bo ZHANG,Shi-qi CHANG,Tian-yi ZHANG. Review on image information hiding methods based on deep learning [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(5): 1497-1515.
[9]	Bin WEN,Yi-fu DING,Chao YANG,Yan-jun SHEN,Hui LI. Self-selected architecture network for traffic sign classification [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(5): 1705-1713.
[10]	Yi-yong PAN,Jia-cong XU,Yi-wen YOU,Yong-jun QUAN. Multi-scale spatial heterogeneity analysis of influencing factors of ride-hailing travel demand [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(5): 1567-1575.
[11]	Jian LI,Huan LIU,Yan-qiu LI,Hai-rui WANG,Lu GUAN,Chang-yi LIAO. Image recognition research on optimizing ResNet-18 model based on THGS algorithm [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(5): 1629-1637.
[12]	Zhen-jiang LI,Li WAN,Shi-rui ZHOU,Chu-qing TAO,Wei WEI. Dynamic estimation of operational risk of tunnel traffic flow based on spatial-temporal Transformer network [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(4): 1336-1345.
[13]	He-shan ZHANG,Meng-wei FAN,Xin TAN,Zhan-ji ZHENG,Li-ming KOU,Jin XU. Dense small object vehicle detection in UAV aerial images using improved YOLOX [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(4): 1307-1318.
[14]	Kai-ming LU,Yan-yan CHEN,Yao TONG,Jian ZHANG,Yong-xing LI,Ying LUO. Data-driven prediction of departure state for tail vehicles in queues at signalized intersections [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(4): 1275-1286.
[15]	Meng-xue ZHAO,Xiang-jiu CHE,Huan XU,Quan-le LIU. A method for generating proposals of medical image based on prior knowledge optimization [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(2): 722-730.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

Comments

Recommended 0

No Suggested Reading articles found!

超参数	数值	超参数	数值
初始学习率	0.012 40	明度增益	0.375 0
学习率因子	0.011 70	旋转角度/（°）	0.000 0
动量	0.980 00	平移幅度	0.090 1
权重衰减	0.000 48	缩放幅度	0.615 0
预热轮次	4.080 00	剪切幅度	0.000 0
预热动量	0.711 00	透视变换	0.000 0
预热偏置学习率	0.108 00	上下翻转概率	0.000 0
边界框损失权重	0.048 30	左右翻转概率	0.500 0
类别损失权重	0.378 00	马赛克增强概率	0.825 0
类别损失权重因子	1.020 00	混合增强概率	0.000 0
目标存在损失权重	0.976 00	复制粘贴增强概率	0.000 0
目标存在损失权重因子	1.070 00	锚框尺度系数	3.690 0
IoU阈值（训练）	0.200 00	色调增益	0.012 4
锚框匹配阈值因子	2.920 00	饱和度增益	0.792 0
Focal Loss 伽马值	0.000 00

CBAM	FasterNet	SIoU	精确率	召回率	平均检测精度	F₁分数
			0.782	0.670	0.699	0.722
√			0.797	0.675	0.705	0.731
√	√		0.806	0.660	0.711	0.726
√	√	√	0.812	0.686	0.715	0.744

模型	精确率	召回率	平均检测精度	参数量	帧率（帧·s^-1）
YOLOv5n	0.683	0.634	0.651	205 728 8	34.44
YOLOv5s	0.781	0.670	0.698	702 772 0	39.84
YOLOv8s	0.731	0.661	0.683	605 728 8	53.01
改进后的YOLOv5	0.812	0.686	0.715	635 776 0	52.08

模型	平均检测精度	精确率	召回率	帧率/（帧·s^-1）
Faster R-CNN	0.507	0.748	0.531	13.62
SSD	0.512	0.676	0.542	35.62
YOLOv5s	0.698	0.781	0.670	39.84
YOLOv8s	0.683	0.731	0.661	53.01
YOLOv11	0.696	0.793	0.686	55.24
改进后的YOLOv5	0.715	0.812	0.686	52.08

Defect recognition of lightweight bridges based on YOLOv5

RICH HTML

PDF (PC)