To solve the problems of multi-modal data and low recognition accuracy in road pavement distress detection, an improved pavement multi-distress recognition algorithm based on the YOLOv8 model enhanced with the non-parametric attention mechanism simAM is proposed. Utilizing the self-owned pavement distress dataset, Res2Net is embedded into the YOLOv8 structure to enhance multi-scale feature extraction capabilities while maintaining similar computational loads. The simAM module is employed to further adjust the weights of feature maps at different scales, improving the detection of targets. Genetic algorithm is used to increase the speed of automatic parameter searching for the model, and image enhancement techniques such as HSV and Mosaic are employed to expand the small sample distresss. Experimental results show that the improved simAM-YOLOv8 algorithm significantly improves accuracy and recall rates for various pavement distresss such as cracks, broken panels, repairs, etc., on asphalt, cement, and other road surfaces. Specifically, the precision rate has increased by 15.3% and the recall rate has increased by 13.1% compared to the original network, demonstrating excellent intelligent recognition performance, and playing an important role in automated detection of highway conditions.
AP(Average-precision)表示平均精度,是主流的目标检测模型的评价指标,通常来说一个越好的分类器,AP值越高。mAP(mean Average precision)是多个类别AP的平均值,mAP的大小一定在[0,1]区间,越大越好。另外,mAP@0.5表示IoU=0.5,计算每一个分类所有图片的AP值,然后所有分类求平均mAP值;mAP@0.5:0.95表示在不同的IoU(从0.5到0.95,不包括0.5,步长为0.05)上的平均mAP值。
LuoHui, YuJun-ying, TuSuo-cheng. Pavement defect detection algorithm based on deep learning[J]. Science Technology and Engineering, 2022, 22(13): 5329-5305.
[3]
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.
[4]
ChengH, WanF, LeiG,et al.GCS-YOLO: a lightweight strawberry disease detection algorithm based on improved YOLOv8[C]∥The 5th International Conference on Frontiers Technology of Information and Computer(ICFTIC), Qiangdao, China, 2023:1071-1076.
ZhangShi-xiang, ZhangHan-cheng, LiXi-zi,et al.Study on multi-objective identification of pavement cracks based on machine vision[J].Journal of Highway and Transportation Rresearch and Development,2021,38(3):30-39.
[7]
陈涵深. 基于深度学习的路面破损检测研究[D]. 杭州: 浙江工业大学信息工程学院, 2020.
[8]
ChenHan-shen.Deep learning for pavement distress detection[D]. Hangzhou: College of Information Engineering,Zhejiang University of Technology, 2020.
WeiChen-hao, YangRui, LiuZhen-bing, et al. YOLOv8 with bi-level routing attention for road scene object detection[J]. Journal of Graphics, 2023, 44(6): 1104-1111.
HouYue, ZhangHui-ting, GaoZhi-wei,et al.Research and comparison of intelligent detection methods of pavement distress based on deep data augmentation[J].Journal of Beijing University of Technology, 2022, 48(6): 622-634.
NiChang-shuang, LiLin, LuoWen-ting, et al. Disease detection of asphalt pavement based on improved YOLOv7[J]. Computer Engineering and Applications, 2023, 59(13): 305-316.
TianPeng, MaoLi. Improved YOLOv8 object detection algorithm for traffic sign target[J]. Computer Engineering and Applications, 2024, 60(8): 202-212.
[17]
YangL, ZhangR Y, LiL, et al. Simam: a simple, parameter-free attention module for convolutional neural networks[C]∥International Conference On Machine Learning, Vienna, Austria, 2021: 11863-11874.
[18]
XieJ, WuZ, ZhuR, et al. Melanoma detection based on swin transformer and SimAM[C]∥IEEE 5th Information Technology, Networking, Electronic and Automation Control Conference(ITNEC), Xi'an, China, 2021: 1517-1521.
[19]
韩强. 面向小目标检测的改进YOLOv8算法研究[D].长春: 吉林大学数学学院, 2024.
[20]
HanQiang. Research on improved YOLOv8 algorithm for small object detection[D].Changchun:School of Mathematics, Jilin University, 2024.
[21]
DuY, PanN, XuZ, et al. Pavement distress detection and classification based on YOLO network[J]. International Journal of Pavement Engineering, 2021, 22(13): 1659-1672.
[22]
JiangP, ErguD, LiuF, et al. A review of YOLO algorithm developments[J]. Procedia Computer Science, 2022, 199: 1066-1073.
ZouYan-yan, CaoYan-fen, ZhangXin-yue, et al. Steel surface defect detection algorithm based on YOLOv8-DSG[J]. Journal of Jilin University(Information Science Edition), 2025, 43(1): 116-125.
[25]
GaoS H, ChengM M, ZhaoK, et al. Res2Net: a new multi-scale backbone architecture[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019, 43(2): 652-662.
[26]
ZhouZ, WangY, FengC, et al. A moving ferromagnetic target localization method based on YOLO-V3[C]∥The 4th International Conference on Intelligent Control, Measurement and Signal Processing (ICMSP),Hangzhou,China,2022: 205-212.
ZongFang, QiHou-cheng, TangMing, et al. Daily travel mode based on GPS data - travel purpose recognition[J]. Journal of Jilin University(Engineering and Technology Edition),2018, 48(5):1374-1379..
[31]
AryaD, MaedaH, GhoshS K, et al.Rdd2022: a multi-national image dataset for automatic road damage detection[DB/OL].[2024-04-26].
[32]
GuoQ, WangY, ZhangY, et al. AIE-YOLO: effective object detection method in extreme driving scenarios via adaptive image enhancement[J]. Science Progress, 2024, 107(3): No. 368504241263165.
[33]
WangH, LiS, ZhangY, et al. Axle surface defect detection based on YOLOv8[C]∥International Conference on Image Processing, Computer Vision and Machine Learning(ICICML), Chengdu, China, 2023:1001-1005.
[34]
YuanJ, WanF, LeiG,et al.FEB-YOLOv8: a steel surface defect detection algorithm based on improved YOLOv8s[C]∥The 3rd International Conference on Computer Science, Electronic Information Engineering and Intelligent Control Technology(CEI),Wuhan, China, 2023: 629-633.