To address the issues of low precision and high model complexity in existing pavement defect detection algorithms, a lightweight automatic detection algorithm for pavement defects based on YOLOv11 was proposed, incorporating a Triplet Attention mechanism (TAD-YOLO11). Firstly, to overcome the limitation of the traditional spatial pyramid pooling-fast (SPPF) module, which solely focused on edge information while neglecting background context, a novel pooling scheme was proposed. This design leveraged adaptive average pooling and adaptive max pooling layers to integrate both global background information and edge details, thereby facilitating effective fusion of multi-scale feature information. Secondly, to enhance cross-dimensional interaction of feature data and compensate for the shortcomings of the original C2PSA module in the channel dimension, the triplet attention mechanism was incorporated into the original module to construct a new lightweight feature fusion module, C2PSA-TA. In the backbone network, the lightweight and efficient upsampling operator DySample was adopted to replace the traditional upsampling algorithm, with triplet attention added to better balance the training data and improve the algorithm’s ability to understand feature information. Finally, testing and validation were conducted on the public datasets RDD2022 and UAV-PDD2023. The results show that compared to YOLOv11n, TAD-YOLO11 exhibits superior performance, specifically, precision improves by 2.5% and 6.9%, respectively, mean average precision increases by 2% and 5.2%, respectively, while the model’s parameter count and computational load are reduced. Additionally, the processing speed is 1.25 times and 2.38 times faster than the baseline model.
LI Rong-hua, CHEN Li-li, WANG Xiang, HE Zhi-xuan.
Automatic Pavement Defect Detection Algorithm Based on Triple Attention Mechanism[J].
Science Technology and Engineering, 2026, 26(8): 3406-3417 DOI:10.12404/j.issn.1671-1815.2501515
HouY, LiQ, ZhangC, et al. The state-of-the-art review on applications of intrusive sensing, image processing techniques, and machine learning methods in pavement monitoring and analysis[J]. Engineering, 2021, 7(6): 845-856.
[2]
DuZ Y, YuanJ, XiaoF P, et al. Application of image technology on pavement distress detection: a review[J]. Measurement, 2021, 184: 109900.
[3]
RatekeT, JustenK A, ChiarellaV F, et al. Passive vision region-based road detection: a literature review[J]. ACM Computing Surveys, 2019, 52(2): 1773-1795.
HouYue, LiQiuhan, ZhangChen, et al. Processing techniques, and machine learning methods in pavement monitoring and analysis[J]. Engineering, 2021, 7(6): 309-334.
LiLi, ZhouRuoling, EGuangxun, et al. Feasibility of fiber optic strain sensor monitoring based on road service conditions[J]. Road, 2024, 69(11): 46-54.
GaoMingxing, JiangZhengfa, ZhangLin, et al. Detection method of pavement cracks in aerial images based on double pyramid network[J]. Laser & Optoelectronics Progress, 2024, 61(22): 122-131.
[12]
ZhangC, LiG, ZhangZ K, et al. AAL-Net: a lightweight detection method for road surface defects based on attention and data augmentation[J]. Applied Sciences-Basel, 2023, 13(3): 239-244.
ZouKaixin, ZhangZijia, SunWei, et al. Improving the road defect image segmentation algorithm of U-Net[J]. Journal of Electronic Measurement and Instrumentation, 2024, 38(8): 15-25.
SunZhaoyun, PeiLili, LiWei, et al. Pavement potting crack detection method based on improved Faster R-CNN[J]. Journal of South China University of Technology (Natural Science Edition), 2020, 48(2): 84-93.
YangWei. Research on Faster R-CNN optimization algorithm for pavement crack identification[D]. Harbin: Harbin University of Science and Technology, 2022.
[22]
GhoshR. On-road vehicle detection in varying weather conditions using Faster R-CNN with several region proposal networks[J]. Multimedia Tools and Applications, 2021, 80(17): 25985-25999.
RenYongjie, WuLipeng. Municipal road disease recognition based on attention mechanism and novel convolutional neural netwoek[J]. Science Technoligy and Engineering, 2024, 24(20): 8663-8672.
YeFamao, ZhangLi, YuanLiao, et al. DB-YOLO: dual backbone YOLOv8 model with feature enhancement fusion for road defect detection[J]. Computer Engineering and Applications, 2024, 60(24): 260-269.
WangXiaoyan, WangXiyu, LiJie, et al. MAS-YOLOv8n road damage detection algorithm from the perspective of drones[J]. Opto-Electronic Engineering, 2024, 51(10): 69-83.
[31]
GuoN, ZhangR, DongQ, et al. EADG-YOLO: road damage detection by improving of YOLOv8n[J]. Journal of Electronic Imaging, 2025, 34(1): 13026.
[32]
ZhangZ, XuZ, ZhangJ. GLE-YOLO: intelligent detection of road linear damage in shadow condition[J]. Signal Image and Video Processing, 2025, 19(3): 1-11.
[33]
HussainM. YOLO-v1 to YOLO-v8, the rise of YOLO and its complementary nature toward digital manufacturing and industrial defect detection[J]. Machines, 2023, 11(7): 1-6.
[34]
VijayakumarA, VairavasundaramS. YOLO-based object detection models: a review and its applications[J]. Multimedia Tools and Applications, 2024, 83(35): 83535-83574.
YeFamao, ZhangLi, YuanLiao, et al. DB-YOLO: dual backbone YOLOv8 model with feature enhancement fusion for road defect detection[J]. Computer Engineering and Applications, 2024, 60(24): 260-269.
[40]
WangJ L, MengR F, HuangY H, et al. Road defect detection based on improved YOLOv8s model[J]. Scientific Reports, 2024, 14(1): 16758.
HeZhixian, CaoLijie. UAVAI-YOLO: dense small target detection algorithm based on UAV aerial images[J]. Chinese Journal of Intelligent Science and Technology, 2024, 6(2): 262-271.
BaiFeng, MaQinglu, ZhaoMin. AC-YOLO for aerial pavement crack detection[J]. Computer Engineering and Applications, 2025, 61(1): 153-164.
[49]
AnwarS, BarnesN. IEEE real image denoising with feature attention[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Seoul: IEEE, 2019: 10729373.
[50]
MisraD, NalamadaT, ArasanipalaiA U, et al. Rotate to attend: convolutional triplet attention module[C]// Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV). New York: IEEE, 2021: 00318.
[51]
WooS H, ParkJ, LeeJ Y, et al. CBAM: convolutional block attention module[C]// Proceedings of the 15th European Conference on Computer Vision (ECCV). Munich: Springer, 2018: 3-19.
[52]
AryaD, MaedaH, GhoshS K, et al. RDD2022: a multi-national image dataset for automatic road damage detection[J]. Geoscience Data Journal, 2024, 11(4): 846-862.
[53]
SelvarajuR R, CogswellM, DasA, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision, 2020, 128(2): 336-359.