Application of YOLO Family Neural Networks for Useful Signals Detection on Eddy Current Rail Defectograms

Improving the level of safety of railway traffic is directly related to the need for prompt detection of structural anomalies of track elements. This task is implemented through regular inspections using non‑destructive testing methods. Among the modern technologies used for this purpose, eddy current flaw detection stands out. The flaw detector generates a multi-channel discrete signal, which is called a defectogram. Defectograms require analysis, that is, the identification of useful signals from a defect or structural elements of the rail. This paper investigates the use of YOLO (You Only Look Once) family convolutional neural networks for automated detection of useful signals in eddy current rail defectograms. The main objective was to evaluate the effectiveness of different transformations of multichannel time‑series data into two‑dimensional images suitable for YOLO processing, and to explore the trade‑off between detection accuracy and computational cost. Four transformation methods are examined: Threshold Transform, based on amplitude comparisons against a twice threshold noise level, Short‑Time Fourier Transform, Continuous Wavelet Transform and Hilbert–Huang Transform. The dataset comprises defectogram fragments of 50 thousand counts with annotated useful signals from three classes (flash butt welds, aluminothermic welds, and bolt joints), split into training, validation, and test sets. YOLO models trained on this data achieved high mean Average Precision scores in useful signals detection for all considered transformation methods. Continuous Wavelet Transform yielded the best scores while the Threshold Transform proved to be the least computationally expensive. Short‑Time Fourier Transform method offered the best balance between precision and recall. Hilbert–Huang Transform showed slightly lower effectiveness. These results demonstrate the suitability of YOLO networks for eddy current defectogram analysis and useful signals detection in general.

1. Kuzmin E. V., Gorbunov O. E., Plotnikov P. O., Tyukin V. A. Finding the Level of Useful Signals on Interpretation of Magnetic and Eddy-Current Defectograms. Automatic Control and Computer Sciences, vol. 52, 2018, pp. 658-666. DOI: 10.3103/S0146411618070179.

2. Быстров Л. Ю., Гладков А. Н., Кузьмин Е. В. Подавление аддитивных периодических низкочастотных помех на вихретоковых дефектограммах. Моделирование и анализ информационных систем, 31(2), с. 164-181. DOI: 10.18255/1818-1015-2024-2-164-181. / Bystrov L.Y., Gladkov A.N., Kuzmin E.V. Suppression of additive periodic low-frequency interference on eddy current defectograms. Modeling and Analysis of Information Systems, 31(2), 2024, pp. 164-181 (In Russian).

3. Chao-Lung Y., Zhi-Xuan C., Chen-Yi Y. Sensor Classification Using Convolutional Neural Network by Encoding Multivariate Time Series as Two-Dimensional Colored Images. Sensors, 20(1), 2019, article no. 168. DOI: 10.3390/s20010168.

4. Terven J., Cordova-Esparza, D. M., Romero-González J. A. A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS. Machine Learning and Knowledge Extraction, 5(4), 2023, pp. 1680-1716. DOI: 10.3390/make5040083.

5. Hussain M. YOLO-v1 to YOLO-v8, the Rise of YOLO and Its Complementary Nature toward Digital Manufacturing and Industrial Defect Detection. Machines, 11(7), 2023, article no. 677. DOI: 10.3390/machines11070677.

6. Wyse L. Audio Spectrogram Representations for Processing with Convolutional Neural Networks. 2017. DOI: 10.48550/arXiv.1706.09559. – Available at: https://arxiv.org/abs/1706.09559.

7. Junior R. F. R. et al. Fault Detection and Diagnosis in Electric Motors Using Convolution Neural Network and Short-Time Fourier Transform. Journal of Vibration Engineering & Technologies, vol. 10, 2022, pp. 1-12. DOI: 10.1007/s42417-022-00501-3.

8. Modupalli V. P. CNN-Based Machine Tool Monitoring with STFT Image Analysis. International Journal for Research in Applied Science and Engineering Technology, 12(5), 2024, pp. 3986-3991. DOI: 10.22214/ijraset.2024.62493.

9. Yu Y., Liu Q., Han B. S., Zhou W. Application of Convolutional Neural Network to Defect Diagnosis of Drill Bits. Applied Sciences, 12(21), 2022, article no. 10799, 13 p. DOI: 10.3390/app122110799.

10. Cheng C., Seo H., Zhao Y. A Novel Pavement Transverse Cracks Detection Model Using WT-CNN and STFT-CNN for Smartphone Data Analysis. International Journal of Pavement Engineering, 23(12), 2021, pp. 1-13. DOI: 10.1080/10298436.2021.1945056.

11. Kang X. et al. Optimal Preprocessing for Joint Detection and Classification of Wireless Communication Signals in Congested Spectrum Using Computer Vision Methods. 2024. DOI: 10.48550/arXiv.2408.06545. – Available at: https://arxiv.org/abs/2408.06545.

12. Liu C., Wang D., Lin Y., Song, S. Research on Online Monitoring of Chatter Based on Continuous Wavelet Transform and Convolutional Neural Network – Vision Transformer (CNN-ViT). Mechanical Sciences, 16(1), 2025, pp. 167-180. DOI: 10.5194/ms-16-167-2025.

13. Lupea I., Lupea M. Continuous Wavelet Transform and CNN for Fault Detection in a Helical Gearbox. Applied Sciences, 15(2), 2025, article no. 950, 23 p. DOI: 10.3390/app15020950.

14. Pałczyński C., Olejnik P. Anomalies Classification in Fan Systems Using Dual-Branch Neural Networks with Continuous Wavelet Transform Layers: An Experimental Study. Information, 16(2), 2025, article no. 71, 13 p. DOI: 10.3390/info16020071.

15. Piedad E. J., Rosario C., Prieto-Araujo E., Bellmunt O. Exploring Wavelet Transformations for Deep Learning-Based Machine Condition Diagnosis. In Proc. of the International Conference on Diagnostics in Electrical Engineering (Diagnostika), 2024, pp. 1-4. DOI: 10.1109/Diagnostika61830.2024.10693895.

16. Siddique M. F, Ahmad Z, Ullah N, Kim J. A Hybrid Deep Learning Approach: Integrating Short-Time Fourier Transform and Continuous Wavelet Transform for Improved Pipeline Leak Detection. Sensors (Basel), 23(19), 2023, article no. 8079. DOI: 10.3390/s23198079.

17. Ch Vidyasagar K. E. et al. Signal to Image Conversion and Convolutional Neural Networks for Physiological Signal Processing: A Review. IEEE Access, vol. 12, 2024, pp. 66726-66764. DOI: 10.1109/ACCESS.2024.3399114.

18. Alvarenga T. A. et al. Detection and Classification System for Rail Surface Defects Based on Eddy Current. Sensors, 21(23), 2021, article no. 7937. DOI: 10.3390/s21237937.

19. Sagar B., Kane P. Intelligent Bearing Fault Diagnosis of Switched Reluctance Motor Using Hilbert Huang Transformation and Region-Based Convolutional Neural Network. The International Journal of Acoustics and Vibration, 29(4), 2024, pp. 431-441. DOI: 10.20855/ijav.2024.29.42072.

20. Srivastava S. et al. Comparative analysis of deep learning image detection algorithms.

21. Journal of Big Data, 8(66), 2021. DOI: 10.1186/s40537-021-00434-w.

22. Available at: https://journalofbigdata.springeropen.com/articles/10.1186/s40537-021 00434-w.

23. Po-Liang Y., Pei-Ling L. Automatic Detection of Crack Echo Signals in Spectrogram Using Yolo. Available at: https://ssrn.com/abstract=4885492. DOI: 10.2139/ssrn.4885492.

24. Vargas V. M. et al. Gramian Angular and Markov Transition Fields Applied to Time Series Ordinal Classification. Advances in Computational Intelligence, 2023, pp. 505-516. DOI: 10.1007/978 3 031 43078-7_41.

25. Zhang Y., Chen X. Motif Difference Field: A Simple and Effective Image Representation of Time Series for Classification. 2020. DOI: 10.48550/arXiv.2001.07582. – Available at: https://arxiv.org/abs/2001.07582.

26. Jun-Hao C., Yun-Cheng T. Dynamic Deep Convolutional Candlestick Learner. 2022. DOI: 10.48550/arXiv.2201.08669. – Available at: https://arxiv.org/abs/2201.08669.

27. Dan-Feng W. et al. Planetary-Gearbox Fault Classification by Convolutional Neural Network and Recurrence Plot. Applied Sciences, 10(3), 2020, article no. 932. DOI: 10.3390/app10030932.

28. Som A. et al. PI-Net: A Deep Learning Approach to Extract Topological Persistence Images. In Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2020, pp. 3639 3648. DOI: 10.1109/CVPRW50498.2020.00425.

29. Samani E. U., Banerjee A. G. Persistent Homology Meets Object Unity: Object Recognition in Clutter. In Proc. of the IEEE Transactions on Robotics, vol. 40, 2024, pp. 886-902. DOI: 10.1109/TRO.2023.3343994.

30. Yang Y., Deng J., Kang D. An Improved Empirical Mode Decomposition by Using Dyadic Masking Signals. Signal, Image and Video Processing, vol. 9, 2015, pp. 1259-1263. DOI: 10.1007/s11760 013 0566-7.

31. Zhaohua W., Jiaxin F., Fang-Li Q., Zhe-Min T. Fast Multidimensional Ensemble Empirical Mode Decomposition for the Analysis of Big Spatio-Temporal Datasets. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 374(2065), 2016, article no. 20150197. DOI: 10.1098/rsta.2015.0197.