YOLOv5 ile Topluluk Öğrenmesine Dayalı Olarak Ray Yüzeyindeki Kusurların Tespiti

Yıl 2023, , 115 - 132, 31.01.2023

Mehmet Sevi , İlhan Aydın , Erhan Akın

https://doi.org/10.47072/demiryolu.1205483

Cited By: 3

Öz

Demiryolu ulaşımı son yıllarda demiryolu hat uzunluğunun artmasıyla beraber kapasitesini arttırmıştır. Hızlı trenlerin gelişmesi de bu duruma katkı sağlamıştır. Yolcu ve yük kapasitesinin artması güvenlik tedbirlerinin önemini daha da arttırmıştır. Demiryolu hatlarının güvenliğini sağlamak için hatların belirli aralıklarla denetlenmesi gerekmektedir. Demiryolu hattı bakımında ray üzerinde bulunan kusurların tespiti son derece önemlidir. Bu çalışmada demiryolu bakımının önemli bir parçası olan ray bileşeni üzerindeki kusurların tespitine odaklanılmıştır. Çalışmada ray üzerinde bulunan kusurları bir nesne tespiti yöntemi olan YOLO ile tespit etme yoluna gidilmiştir. Farklı YOLO modelleri için topluluk öğrenmesine dayalı bir yöntem önerilmiştir. Deney sonuçları, 8 farklı kusur içeren veri seti üzerinde bütün sınıfları içeren tespit oranının %80’in üzerinde olduğunu göstermiştir.

Anahtar Kelimeler

Demiryolu, Ray bağlantı elemanları, Kusur tespiti, Bilgisayarlı görme, YOLO, Derin öğrenme algoritmaları., Railway, Rail fasteners, Defect detection, Computer vision, Deep learning algorithms

Destekleyen Kurum

TÜBİTAK

Proje Numarası

120E097

Teşekkür

Bu çalışma, 120E097 numaralı TÜBİTAK projesi tarafından desteklenmiştir.

Detection of Rail Surface Defects Based on Ensemble Learning of YOLOv5

Öz

Railway transportation has increased its capacity with the increase in railway line length in recent years. The development of high-speed trains also contributed to this situation. The increase in passenger and cargo capacity has further increased the importance of security measures. In order to ensure the safety of the railway line, it is necessary to inspect the line at certain intervals. Detection of defects on the rail is extremely important in the maintenance of the railway line. This study focuses on the detection of defects on the rail component, which is an important part of railway maintenance. In the study, it was tried to detect the defects on the rail with YOLO, which is an object detection method. In the study, it has been shown that model ensembling gives better results than YOLO models that validate alone. A method based on ensemble learning is proposed for different YOLO models. Experiment results showed that the detection rate including all classes on the data set containing 8 different defects was over 80%.

Anahtar Kelimeler

Demiryolu, Ray bağlantı elemanları, Kusur tespiti, Bilgisayarlı görme, YOLO, Derin öğrenme algoritmaları., Railway, Rail fasteners, Defect detection, Computer vision, Deep learning algorithms

Proje Numarası

120E097

Kaynakça

• [1] O. Yaman, Demiryolu Rayları İçin Gerçek Zamanlı Bulanık Otomata ile Görme Tabanlı Arıza Teşhis Sisteminin Geliştirilmesi, PhD Thesis, Firat University, 2018.
• [2] C. Taştimur, M. Karaköse, E. Akın and İ. Aydın, "Rail defect detection with real time image processing technique," 2016 IEEE 14th International Conference on Industrial Informatics (INDIN), 2016, pp. 411-415, doi: 10.1109/INDIN.2016.7819194.
• [3] D. Barke, & W. K. Chiu, “Structural Health Monitoring in the Railway Industry: A Revie,” Structural Health Monitoring, 2005, 4(1), 81–93.
• [4] D. Çetintaş, T. Tuncer, “Determining the type of document read using eye movement properties by hybrid CNN method,” Traitement du Signal, vol. 39, No. 4, pp. 1099-1108, 2022, doi: 10.18280/ts.390402.
• [5] H. Feng, Z. Jiang, F. Xie, P. Yang, J. Shi and L. Chen, "Automatic Fastener Classification and Defect Detection in Vision-Based Railway Inspection Systems," in IEEE Transactions on Instrumentation and Measurement, vol. 63, no. 4, pp. 877-888, 2014.
• [6] F. Guo, Y. Qian, and Y. Shi, “Real-time railroad track components inspection based on the improved yolov4 framework,” Automation in Construction, 2021.
• [7] D. Zheng et al., “A Defect Detection Method for Rail Surface and Fasteners Based on Deep Convolutional Neural Network,” Computational Intelligence and Neuroscience, vol. 2021, p. e2565500, Aug. 2021, doi: 10.1155/2021/2565500.
• [8] Y. Xia, F. Xie and Z. Jiang, "Broken Railway Fastener Detection Based on Adaboost Algorithm," 2010 International Conference on Optoelectronics and Image Processing, 2010, pp. 313-316, doi: 10.1109/ICOIP.2010.303.
• [9] H. Fan, P. C. Cosman, Y. Hou and B. Li, "High-Speed Railway Fastener Detection Based on a Line Local Binary Pattern," in IEEE Signal Processing Letters, vol. 25, no. 6, pp. 788-792, June 2018, doi: 10.1109/LSP.2018.2825947.
• [10] L. Shang, Q. Yang, J. Wang, S. Li and W. Lei, "Detection of rail surface defects based on CNN image recognition and classification," 2018 20th International Conference on Advanced Communication Technology (ICACT), pp. 45-51, 2018.
• [11] H. Yu et al., "A Coarse-to-Fine Model for Rail Surface Defect Detection," in IEEE Transactions on Instrumentation and Measurement, vol. 68, no. 3, pp. 656-666, March 2019, doi: 10.1109/TIM.2018.2853958.
• [12] S. Yanan, Z. Hui, L. Li and Z. Hang, "Rail Surface Defect Detection Method Based on YOLOv3 Deep Learning Networks," 2018 Chinese Automation Congress (CAC), 2018, pp. 1563-1568, doi: 10.1109/CAC.2018.8623082.
• [13] Li H, Wang F, Liu J, Song H, Hou Z, et al. (2022) Ensemble model for rail surface defects detection. PLOS ONE 17(5), doi: 10.1371/journal.pone.0268518.
• [14] Shorten, C., Khoshgoftaar, T.M. A survey on Image Data Augmentation for Deep Learning. J Big Data 6, 60 (2019). https://doi.org/10.1186/s40537-019-0197-0.
• [15] A. Akdağ, Derin öğrenme algoritmaları kullanılarak gerçek zamanlı silah tanıma uygulaması, Master Thesis, Necmettin Erbakan University, 2017.
• [16] J. Redmon, S. Divvala, R. Girshick and A. Farhadi, "You Only Look Once: Unified, Real-Time Object Detection," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779-788, doi: 10.1109/CVPR.2016.91.
• [17] E. Güçlü , İ. Aydın , K. Şahbaz , E. Akın ve M. Karaköse , "Demiryolu Bağlantı Elemanlarında Bulunan Kusurların YOLOv4 ve Bulanık Mantık Kullanarak Tespiti", Demiryolu Mühendisliği, vol. 14, pp. 249-262, 2021, doi:10.47072/demiryolu.939830.
• [18] Z. Zakria, J. Deng, R. Kumar, M. S. Khokhar, J. Cai and J. Kumar, "Multiscale and Direction Target Detecting in Remote Sensing Images via Modified YOLO-v4," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 15, pp. 1039-1048, 2022, doi: 10.1109/JSTARS.2022.3140776.
• [19] YOLOv5 Custom Training. https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data (accessed August. 8, 2022).
• [20] E. Güney, Sürücü asistan sistemleri için mobil gpu tabanlı gerçek zamanlı durum analizi ve tespit uygulamaları, Master Thesis, Sakarya University, 2021.
• [21] R. Xu, H. Lin, K. Lu, L. Cao, Y. Liu, “A Forest Fire Detection System Based on Ensemble Learning,” Forests. 2021; 12(2):217, doi: 10.3390/f12020217
• [22] V. Kotu, B. Deshpande, “Chapter 2 - Data Science Process.” ScienceDirect, Morgan Kaufmann, 2019.
• [23] G. Conley, S. C. Zinn, T. Hanson, K. McDonald, N. Beck, and H. Wen, “Using a deep learning model to quantify trash accumulation for cleaner urban stormwater,” Computers, Environment and Urban Systems, 2022.
• [24] R. Padilla, S. L. Netto and E. A. B. da Silva, "A Survey on Performance Metrics for Object-Detection Algorithms," 2020 International Conference on Systems, Signals and Image Processing (IWSSIP), 2020, pp. 237-242, doi: 10.1109/IWSSIP48289.2020.9145130.
• [25] H. Yu et al., "A Coarse-to-Fine Model for Rail Surface Defect Detection," in IEEE Transactions on Instrumentation and Measurement, vol. 68, no. 3, pp. 656-666, March 2019, doi: 10.1109/TIM.2018.2853958
• [26] C. Zhang, X. Hu, J. He, N. Hou, "Yolov4 High-Speed Train Wheelset Tread Defect Detection System Based on Multiscale Feature Fusion", Journal of Advanced Transportation, vol. 2022, Article ID 1172654, 13 pages, 2022, doi: 10.1155/2022/1172654.