(49-3) 10 * << * >> * Russian * English * Content * All Issues

Research of neural network algorithms for recognizing railway infrastructure objects in video images
E.V. Medvedeva 1, A.A. Perevoshchikova 1

1 Vyatka State University,
Moskovskaya Str. 36, Kirov, 610000, Russia

 PDF, 3797 kB

DOI: 10.18287/2412-6179-CO-1563

Pages: 443-450.

Full text of article: Russian language.

Abstract:
The article describes the development of two neural network algorithms for recognizing objects of the railway infrastructure in video images. Both algorithms are aimed at improving railway traffic safety. One algorithm detects foreign objects on railway tracks and objects relating to the railway infrastructure. The other algorithm implements the semantic segmentation of main and auxiliary railway tracks, as well as trains within the visible range of the locomotive. The algorithms are implemented based on convolutional neural networks (CNN) YOLO and U-Net. The CNN is trained and tested using the image database of the Research Institute of Information, Automation and Communications in Railway Transport. The experimental studies conducted are aimed at increasing the efficiency of algorithms for object detection and segmentation through the use of data augmentation methods and additional preprocessing, as well as selecting an architecture and optimal network hyperparameters. The detection algorithm works in real time, achieving an average accuracy of 64% for 11 object classes according to the mAP metric. The operating speed of the semantic segmentation algorithm is 5 frames/s, the average accuracy for three classes of objects according to the IoU metric is 92%.

Keywords:
object detection, semantic segmentation, railway infrastructure objects, railway traffic safety, machine vision systems, neural network algorithms.

Citation:
Medvedeva EV, Perevoshchikova AA. Research of neural network algorithms for recognizing railway infrastructure objects in video images. Computer Optics 2025; 49(3): 443-450. DOI: 10.18287/2412-6179-CO-1563.

References:
  1. Popov PA. Application of artificial intelligence technologies for railway transport [In Russian]. Bulletin of the Institute for Problems of Natural Monopolies: Railway Engineering 2024; 1(65): 38-41.
  2. Ozerov AV, Okhotnikov AL. Computer vision as part of advanced train protection system [In Russian]. Intelligent transport systems. Materials of the II International Scientific and Practical Conference 2023: 620-625. DOI 10.30932/9785002182794-2023-620-625.
  3. Chen P, Zhang Z, Huang Y, Dai L, Xu F, Hu H. Railway obstacle intrusion warning mechanism integrating YOLO-based detection and risk assessment. J Ind Inf Integr 2024; 38: 100571. DOI: 10.1016/j.jii.2024.100571.
  4. He D, Ren R, Li K, Zou Z, Ma R, Qin Y, Yang W, Urban rail transit obstacle detection based on Improved R-CNN. Measurement 2022; 196: 111277. DOI: 10.1016/j.measurement.2022.111277.
  5. Medvedeva EV, Kurbatova EE. Image segmentation based on two-dimensional markov chains. In Book: Favorskaya MN, Jain LC, eds. Computer vision in Control Systems-2. Innovations in practice. Cham, Switzerland: Springer International Publishing Switzerland; 2015: 277-295. DOI: 10.1007/978-3-319-11430-9_11.
  6. Silar Z, Dobrovolny M. Utilization of directional properties of optical flow for railway crossing occupancy monitoring. 2013 Int Conf on IT Convergence and Security (ICITCS), 2013, pp. 1-4. DOI: 10.1109/ICITCS.2013.6717896.
  7. Bagaev S, Medvedeva E. Experimental assessment of accuracy of multiclass segmentation of objects from satellite images based on a modified convolutional neural network U-net [In Russian]. Current Problems in Remote Sensing of the Earth from Space 2021; 18(6): 35-45. DOI: 10.21046/2070-7401-2021-18-6-35-45.
  8. Bagaev S. Medvedeva E. Segmentation of satellite images of the earth's surface using neural network technologies. 28th Conf of Open Innovations Association (FRUCT) 2021: 15-21. DOI: 10.23919/FRUCT50888.2021.9347653.
  9. Orlov SP, Efimushkin NA, Efimushkina NV. Deep neural network for diagnosing railway track elements [In Russian]. Vestnik of Samara State Technical University. Technical Sciences Series 2022; 30(1): 63-72. DOI: 10.14498/tech.2022.1.4.
  10. Minakov VA, Fomenko VK. Machine vision technology on locomotives for identifying track signals [In Russian]. World of Transport 2019; 17(6): 62-72. DOI: 10.30932/1992-3252-2019-17-62-72.
  11. Rahman F, Ahmed Md, Hasan Md, Jahan N. Real-time obstacle detection over railway track using deep neural networks. Procedia Comput Sci 2022; 215: 289-298. DOI: 10.1016/j.procs.2022.12.031.
  12. Chen Z, Yang J, Chen L, Feng Z, Jia L. Efficient railway track region segmentation algorithm based on lightweight neural network and cross-fusion decoder. Autom Constr 2023; 155: 105069. DOI: 10.1016/j.autcon.2023.105069.
  13. Zeng Y, Wang X, Zhao H, Jin Y, Giannopoulos GA, Lia Y. Image fusion methods in high-speed railway scenes: A survey. High-Speed Railway 2023; 1(2): 87-91. DOI: 10.1016/j.hspr.2023.01.001.
  14. Mahendrakar T, Ekblad A, Fischer N, White R, Wilde M, Kish B, Silver I. Performance study of YOLOv5 and Faster R-CNN for autonomous navigation around non-cooperative targets. 2022 IEEE Aerospace Conf (AERO) 2022: 1-12. DOI: 10.1109/AERO53065.2022.9843537.
  15. Rahmaniar W, Hernawan A. Real-time human detection using deep learning on embedded platforms: A review. J Robot Control 2021; 2(6): 462-468. DOI: 10.18196/jrc.26123.
  16. Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu C-Y, Berg A-C. SSD: Single shot multibox detector. In Book: Leibe B, Matas J, Sebe N, Welling M, eds. Computer Vision – ECCV 2016. 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I. Cham, Switzerland: Springer International Publishing AG; 2016: 21-37. DOI: 10.1007/978-3-319-46448-0_2.
  17. Yang G, Wang Z, Zhuang S. PFF-FPN: A parallel feature fusion module based on FPN in Pedestrian detection. 2021 Int Conf on Computer Engineering and Artificial Intelligence (ICCEAI) 2021: 377-381. DOI: 10.1109/ICCEAI52939.2021.00075.
  18. Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. ICML'15: Proc 32nd Int Conf on International Conference on Machine Learning 2015; 37: 448-456.
  19. Zhang H, Jiang Z, Zheng G, Yao X. Semantic segmentation of UAV remote sensing images based on improved U-Net. 2023 8th Int Conf on Intelligent Computing and Signal Processing (ICSP) 2023: 1735-1740. DOI: 10.1109/ICSP58490.2023.10248526.
  20. He K, Gkioxari G, Dollár P, Girshick R. Mask R-CNN. 2017 IEEE International Conference on Computer Vision (ICCV) 2017: 2961-2969. DOI: 10.1109/ICCV.2017.322.
  21. Khrueakhrai S, Srinonchat J. Railway track detection based on SegNet deep learning. TENCON 2023 – 2023 IEEE Region 10 Conf (TENCON) 2023: 409-413. DOI: 10.1109/TENCON58879.2023.10322378.
  22. Seeli D, Thanammal KK. Brain disease detection using PSPNet for multiclass segmentation. 2nd Int Conf on Applied Artificial Intelligence and Computing (ICAAIC) 2023: 1030-1034. DOI: 10.1109/ICAAIC56838.2023.10140366.
  23. Zhang Y, Zhang Y, Zhang Q. Semantic segmentation of traffic scene based on DeepLabv3+ and attention mechanism. 3rd Int Conf on Neural Networks, Information and Communication Engineering (NNICE) 2023: 542-547. DOI: 10.1109/NNICE58320.2023.10105805.

© 2009, IPSI RAS
151, Molodogvardeiskaya str., Samara, 443001, Russia; E-mail: journal@computeroptics.ru ; Tel: +7 (846) 242-41-24 (Executive secretary), +7 (846) 332-56-22 (Issuing editor), Fax: +7 (846) 332-56-20