Research on Pavement Crack Identification Method Based on Improved YOLOv11

Xuanyu Fang

doi:10.54097/q26apx03

Authors

Xuanyu Fang

DOI:

https://doi.org/10.54097/q26apx03

Keywords:

Pavement Crack Detection, Dynamic Convolution, Attention Mechanism, Loss Function

Abstract

Pavement cracks are the most predominant form of distress in asphalt pavements; therefore, pavement crack detection constitutes a critical component of pavement maintenance. To address the issues of low recognition accuracy, false detection, and missed detection in highway pavement crack identification, this paper proposes a pavement crack detection model based on YOLOv11.First, Ghost convolution and DynamicConv are integrated into the C3K2 module to reduce computational load and enhance feature extraction capabilities. Second, the CGA attention mechanism is introduced in combination with C2PSA to strengthen feature fusion and contextual information extraction. Finally, the Adaptive Threshold Focal Loss function is employed to address the issue of class imbalance and improve the detection capability for difficult samples. Experiments conducted on a self-constructed dataset demonstrate that the improved model outperforms existing methods in both accuracy and efficiency. The proposed method provides an efficient and accurate solution for pavement crack detection.

Downloads

Download data is not yet available.

References

[1] Liu G, Wu X, Dai F, et al. Crack-MsCGA: a deep learning network with multi-scale attention for pavement crack detection[J]. Sensors, 2025, 25(8): 24-46.

[2] Geng H, Liu Z, Wang Y, et al. SDFC-YOLO: a YOLO-based model with selective dynamic feature compensation for pavement distress detection[J]. IEEE Transactions on Intelligent Transportation Systems, 2025, 26(2): 1842-1856.

[3] Zhang J, Sun S, Song W, et al. Automated pavement distress detection based on convolutional neural network[J]. IEEE Access, 2024, 12: 105055-105068.

[4] Han K, Wang Y, Tian Q, et al. GhostNet: more features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 1580-1589.

[5] Chen Y, Dai X, Liu M, et al. Dynamic convolution: attention over convolution kernels[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 11030-11039.

[6] Zhao Y, Miao J, Li Z, et al. CGA-ViT: channel-guided additive attention for efficient vision recognition[J]. Applied Sciences, 2026, 16(4): 1740.

[7] Yang B, Zhang X, Zhang J, et al. EFLNet: enhancing feature learning network for infrared small target detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 1-11.

[8] Sunkara R, Luo T. No more strided convolutions or pooling: a new CNN building block for low-resolution images and small objects[C]//Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Cham: Springer, 2022: 443-459.

[9] Zong F, Zhao K, Jiang S, et al. Detecting highway pavement diseases by developing an improved YOLOv5 algorithm[C]// International Conference on Artificial Intelligence and Autonomous Transportation. Singapore: Springer, 2024: 142-151.

[10] Li Z, Peng Y, Liu M, et al. Asphalt pavement raveling identification based on machine learning[C]//Fourth International Conference on Image Processing and Intelligent Control. Bellingham: SPIE, 2024, 13250: 214-219.

[11] Huang Z, Chen X, Liu H, et al. Pavement diseases detection using improved YOLOv5[C]//2023 IEEE International Conference on Mechatronics and Automation. Piscataway: IEEE, 2023: 1786-1791.

[12] Liang Z, Di R, Tan F, et al. Fig-YOLO: an improved YOLOv11-based fig detection algorithm for complex environments[J]. Foods, 2025, 14(23): 41-54.