RSD-YOLO: An Improved YOLOv11n-Based Algorithm for Railway Surface Defect Detection
DOI:
https://doi.org/10.63313/AJET.9057Keywords:
RSD-YOLO, YOLOv11n, Railway Surface Defect Detection, P2 Detection Head, C2f, SPPFAbstract
Railway surface defect detection is a critical component of automated track maintenance and safety assurance. To improve the detection capability of lightweight YOLOv11n on rail surface images, this paper proposes RSD-YOLO, a conservative structural enhancement that adds a P2 high-resolution detection branch, replaces selected high-level C3k2 blocks with C2f modules, and introduces P2-SPPF context enhancement. The resulting detector changes the original three-scale P3/P4/P5 output to a four-scale P2/P3/P4/P5 output. Experiments on a four-class rail surface defect dataset show that RSD-YOLO achieves 0.7966 [email protected] and 0.5616 [email protected]:0.95 on the test set, improving YOLOv11n by 8.05 and 7.24 percentage points, respectively.
References
[1] Su, B., Chen, H., & Zhou, Z. (2022). BAF-detector: An efficient CNN-based detector for photovoltaic cell defect detection. IEEE Transactions on Industrial Electronics, 69(3), 3161-3171.
[2] Zhang, Y., Zhang, H., Huang, Q., Han, Y., & Zhao, M. (2024). DsP-YOLO: An anchor-free network with DsPAN for small object detection of multiscale defects. Expert Systems with Applications, 238, 122669.
[3] Hao, X., et al. (2024). An improved YOLOv8 algorithm for rail surface defect detection. IEEE Access, 12, 45210-45222.
[4] Liu, J., Shen, X., Wang, J., Jiang, L., & Zhang, H. (2023). An intelligent defect detection approach based on cascade attention network under complex magnetic flux leakage signals. IEEE Transactions on Industrial Electronics, 71(2), 2056-2066.
[5] Kim, H., et al. (2022). Defect detection of subway tunnels using advanced U-Net network. Sensors, 22(6), 2335.
[6] Li, X., Pu, X., Ling, W. et al. YOLO-SAM an end-to-end framework for efficient real time object detection and segmentation. Sci Rep 15, 40854 (2025). https://doi.org/10.1038/s41598-025-24576-6
[7] Han Y, Qi K, Zheng J, et al. Lightweight Cattle Facial Recognition Method Based on Improved YOLOv11. Smart Agriculture, 2025, 7(3): 173-184. https://doi.org/10.12133/j.smartag.SA202502010
[8] Xu, X. et al. Exploring image enhancement for salient object detection in low light images. ACM Trans. Multimed. Comput. Commun. Appl. 17, 1–19 (2021).
[9] Gao, L., Zhang, J., Yang, C., & Zhou, Y. (2022). Cas-VSwin transformer: A variant swin transformer for surface-defect detection. Computers in Industry, 140, 103689.
[10] Wang Yin, Wang Lide, Qiu Ji. Real-time Enhancement Algorithm for Orbital Dark Light Environment Based on DenseNet Structure[J]. Journal of Southwest Jiaotong University, 2022, 57(6): 1349-1357. doi: 10.3969/j.issn.0258-2724.20210199
[11] Xue R ,Duan J ,Du Z .MPE-DETR: A multiscale pyramid enhancement network for object detection in low-light images[J].Image and Vision Computing,2024,150105202-105202.DOI:10.1016/J.IMAVIS.2024.105202.
[12] Dong, X., Zhang, C., Wang, J., Chen, Y., & Wang, D. (2024). Real-time detection of surface cracking defects for large-sized stamped parts. Computers in Industry, 158, 104105.
[13] Chen, Z., Yang, C., Li, Q., Zhao, F., Zha, Z. J., & Wu, F. (2024). A Dual-Task Branch Alignment Detector for Object Detection. IEEE Transactions on Image Processing, 33, 1642-1655.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 by author(s) and Erytis Publishing Limited.
This work is licensed under a Creative Commons Attribution 4.0 International License.
How to Cite
