Autors: Shaikh M.Z., Jatoi S., Baro E.N., Abro B., Manolova, A. H., Chowdhry B.S.
Title: Attention is All Railway Wheels Need in the 21st Century: Revolutionizing Real-Time Defect Detection with NCRA-YOLO
Keywords: Computer Vision, Deep Learning, NCRA-YOLO, Railway Safety, Wheelset Defect Detection

Abstract: The safety and operational efficiency of railway infrastructural heavily relies on the condition of railway wheelsets and effective defect detection. Traditional inspection methods such as visual inspection and ultrasonic testing show limitations in terms of precision, operational efficiency and scalability. In this study, a novel approach is presented for real-time wheelset defect detection using state-of-the-art deep learning model: Neural Compact Refined Architecture-You Look Only Once model (NCRA-YOLO). A large dataset consists of 7487 high-resolution annotated images of different wheelset. The dataset used for training and testing utilized advanced data augmentation techniques to enhance performance. The model demonstrated superior accuracy of 100% in fracture and 83% in spot class. This model is the most effective solution with only 1.3 million parameters and 3.8 GFLOPS that proves its effectiveness in real time defect detection. The model achieved a precision of 94.7% and a mean average precision at IoU thresholds of 50% to 95% (mAP50:95) of 96.1% during cross validation, making it as the most effective model for generalizing to unseen data.

References

  1. P. Li, R. Xue, S. Shao, Y. Zhu, and Y. Liu, "Current state and predicted technological trends in global railway intelligent digital transformation, " Railway Sciences, vol. 2, no. 4, pp. 397-412, Nov. 2023, doi: 10. 1108/RS-10-2023-0036.
  2. A. Bracciali, "Railway Wheelsets: History, Research and Developments, " International Journal of Railway Technology, vol. 5, no. 1, pp. 23-52, 2016, doi: 10. 4203/IJRT. 5. 1. 2.
  3. N. Chaar, "Wheelset Structural Flexibility and Track Flexibility in Vehicle-Track Dynamic Interaction, " 2007.
  4. T. A. Zucarelli, M. A. Vieira, L. A. Moreira Filho, D. A. P. Reis, and L. Reis, "Failure analysis in railway wheels, " Procedia Structural Integrity, vol. 1, pp. 212-217, Jan. 2016, doi: 10. 1016/J. PROSTR. 2016. 02. 029.
  5. M. Z. Shaikh, S. Jatoi, E. N. Baro, B. Das, S. Hussain, and B. S. Chowdhry, "FaultSeg: A Dataset for Train Wheel Defect Detection, " Scientific Data 2025 12: 1, vol. 12, no. 1, pp. 1-12, Feb. 2025, doi: 10. 1038/s41597-025-04557-0.
  6. A. Alemi, F. Corman, and G. Lodewijks, "Condition monitoring approaches for the detection of railway wheel defects, " http://dx. doi. org/10. 1177/0954409716656218, vol. 231, no. 8, pp. 961-981, Jun. 2016, doi: 10. 1177/0954409716656218.
  7. J. S. Jwo, C. S. Lin, C. H. Lee, L. Zhang, and S. M. Huang, "Intelligent System for Railway Wheelset Press-Fit Inspection Using Deep Learning, " Applied Sciences 2021, Vol. 11, Page 8243, vol. 11, no. 17, p. 8243, Sep. 2021, doi: 10. 3390/APP11178243.
  8. "Preventive NDT methods for railway line bolts, train axels and wheels. " Accessed: Feb. 24, 2025. [Online]. Available: https://www. screeningeagle. com/en/inspection/ndt-for-railway-line-bolts-train-axels-wheels
  9. R. Ferdousi, F. Laamarti, C. Yang, and A. El Saddik, "A Reusable AI-Enabled Defect Detection System for Railway Using Ensembled CNN, " Nov. 2023, Accessed: Feb. 24, 2025. [Online]. Available: https://arxiv. org/abs/2311. 14824v1
  10. Abro, B., Jatoi, S., Shaikh, M. Z., Baro, E. N., Chowdhry, B. S., & Milanova, M. (2025). Towards Smarter Road Maintenance: YOLOv7-Seg for Real-Time Detection of Surface Defects. Lecture Notes in Computer Science, 15618 LNCS, 36-54. https://doi. org/10. 1007/978-3-031-88220-3_3
  11. Jatoi, S., Abro, B., Mushtaq, N., Shah Syed, A. A., Narejo, S., Rathi, M., & Maryam, N. (2024). From Detection to Diagnosis: Elevating Track Fault Identification with Transfer Learning. 6th International Conference on Robotics and Automation in Industry, ICRAI 2024. https://doi. org/10. 1109/ICRAI62391. 2024. 10894696
  12. Shaikh, M. Z., Baro, E. N., & Jatoi, S. (2025). A Contribution to Reliable Rail Transport: AI-Powered Real-Time Wheel Defect Detection. https://doi. org/10. 21203/RS. 3. RS-6615019/V1
  13. M. Z. Shaikh, S. Jatoi, E. N. Baro, B. Das, S. Hussain, and B. S. Chowdhry, "FaultSeg: A Dataset for Train Wheel Defect Detection, " Scientific Data 2025 12: 1, vol. 12, no. 1, pp. 1-12, Feb. 2025, doi: 10. 1038/s41597-025-04557-0.
  14. M. Z. Shaikh et al., "Design and Development of a Wayside AI-Assisted Vision System for Online Train Wheel Inspection, " Engineering Reports, vol. 7, no. 1, p. e13027, Jan. 2025, doi: 10. 1002/ENG2. 13027.
  15. M. Z. Shaikh, Z. Ahmed, E. N. Baro, S. Hussain, and M. Milanova, "Deep learning based identification and tracking of railway bogie parts, " Alexandria Engineering Journal, vol. 107, pp. 533-546, Nov. 2024, doi: 10. 1016/J. AEJ. 2024. 07. 064.
  16. Z. Zhang, S. Shao, and Z. Gao, "A novel method on wheelsets geometric parameters on line based on image processing, " 2010 International Conference on Measuring Technology and Mechatronics Automation, ICMTMA 2010, vol. 1, pp. 257-260, 2010, doi: 10. 1109/ICMTMA. 2010. 58.
  17. M. Z. Shaikh et al., "State-of-the-Art Wayside Condition Monitoring Systems for Railway Wheels: A Comprehensive Review, " IEEE Access, vol. 11, pp. 13257-13279, 2023, doi: 10. 1109/ACCESS. 2023. 3240167.
  18. C. Peng et al., "Automatic railway wheelset inspection system by using ultrasonic technique, " https://doi. org/10. 1117/12. 904998, vol. 8321, pp. 596-601, Nov. 2011, doi: 10. 1117/12. 904998.
  19. M. Z. Shaikh, E. N. Baro, Y. A. Baloch, T. Hussain, M. A. Uqaili, and B. S. Chowdhry, "AI-Powered Real-Time Detection of Wheel Defects in Railways Using Yolov8, " 2024 Global Conference on Wireless and Optical Technologies (GCWOT), pp. 1-7, Sep. 2024, doi: 10. 1109/GCWOT63882. 2024. 10805632.
  20. M. Z. Shaikh, S. Ali, S. Ali, E. N. Baro, Y. A. Baloch, and B. S. Chowdhry, "Predictive Maintenance in Urban Railway Systems Using Machine Learning Models, " 2024 Global Conference on Wireless and Optical Technologies (GCWOT), pp. 1-5, Sep. 2024, doi: 10. 1109/GCWOT63882. 2024. 10805699.
  21. Y. Song, Z. Ji, X. Guo, Y. Hsu, Q. Feng, and S. Yin, "A comprehensive laser image dataset for real-time measurement of wheelset geometric parameters, " Scientific Data 2024 11: 1, vol. 11, no. 1, pp. 1-9, May 2024, doi: 10. 1038/s41597-024-03288-y.
  22. E. Bernal, M. Spiryagin, and C. Cole, "Ultra-Low Power Sensor Node for On-Board Railway Wagon Monitoring, " IEEE Sens J, vol. 20, no. 24, pp. 15185-15192, Dec. 2020, doi: 10. 1109/JSEN. 2020. 3011132.
  23. Y. Q. Ni and Q. H. Zhang, "A Bayesian machine learning approach for online detection of railway wheel defects using track-side monitoring, " Struct Health Monit, vol. 20, no. 4, pp. 1536-1550, Jul. 2021, doi: 10. 1177/1475921720921772/ASSET/IMAGES/10. 1177_1475921720921772-IMG1. PNG.
  24. J. Sresakoolchai and S. Kaewunruen, "Wheel flat detection and severity classification using deep learning techniques, " Insight: Non-Destructive Testing and Condition Monitoring, vol. 63, no. 7, pp. 393-402, Jul. 2021, doi: 10. 1784/INSI. 2021. 63. 7. 393.
  25. R. Paul, M. Neeraj, N. Sagar, V. Nair, and Y. Ps, "A Swarm-based AI Aided Wheel Bot System to Detect Cracks in Railway Tracks, " 6th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2022-Proceedings, pp. 678-685, 2022, doi: 10. 1109/I-SMAC55078. 2022. 9987326.

Issue

International Symposium on Wireless Personal Multimedia Communications, WPMC, 2026, Bulgaria, https://doi.org/10.1109/WPMC67460.2025.11351276

Copyright IEEE

Вид: публикация в международен форум, публикация в реферирано издание, индексирана в Scopus