Russian Federation
Russian Federation
Russian Federation
UDC 004.93
UDC 656.2
this article addresses the problem of selecting and substantiating neural network architectures for computer vision systems in railway transport. In the context of digital transformation within the industry, the development of reliable intelligent decision support systems capable of real-time operation has become a critical priority. This study presents a comparative analysis of the effectiveness of modern convolutional neural networks and emerging transformer-based architectures for the semantic segmentation of key railway infrastructure objects. The research is based on a specialized dataset comprising 8,203 images captured by onboard locomotive cameras under diverse weather and lighting conditions. Five artificial intelligence models were evaluated: U-Net, U-Net++, DeepLabV3+, MAnet, and SegFormer. Performance assessment was conducted using criteria including the Mean Intersection over Union (mIoU) metric, inference speed, and analysis of Type I (false alarm) and Type II (missed detection) error probabilities. Experimental results demonstrate that the U-Net++ architecture with a DenseNet-121 encoder delivers the most balanced performance, minimizing critical missed detections while maintaining high localization accuracy. The findings and proposed evaluation criteria provide a structured framework for the substantiated selection of machine vision algorithms in the design of autonomous control systems and intelligent decision support systems for operators of track maintenance machinery.
semantic segmentation, convolutional neural networks, transformers, computer vision, railway infrastructure, artificial intelligence
1. Zelova M. I., Komarov A. V. O primenenii kompleksa avtomatizirovannogo upravleniya dvizheniem poezdov v usloviyah vysokoy intensivnosti dvizheniya // Sovremennye tehnologii. Sistemnyy analiz. Modelirovanie. 2022. № 4 (76). S. 92–100. DOI:https://doi.org/10.26731/1813-9108.2022.4(76).92-100. EDN WFQNPA
2. Robotizaciya na zheleznodorozhnom transporte / A. I. Dolgiy [i dr.] // Intellektual'nyy transport. 2025. T. 9, № 3 (35). S. 4–32. EDN JCBFIE
3. Shebe H., Shubinskiy I. B., Rozenberg E. N. Razlichnye podhody k avtonomnomu vozhdeniyu dlya zheleznyh dorog // Nadezhnost'. 2025. T. 25, № 1. S. 4–10. DOI:https://doi.org/10.21683/1729-2646-2025-25-1-4-10. EDN DTBFTK
4. Kornev D. A. Modelirovanie setey peredachi dannyh v intellektual'noy sisteme upravleniya dvizheniem poezdov // Izvestiya Transsiba. 2023. № 3 (55). S. 141– 154. EDN QICJOC
5. Osipov D. V. Mashinnoe zrenie na transporte: perspektivy i primenenie // Ekonomika zheleznyh dorog. 2023. № 8. S. 71–75. EDN IAHQWG
6. Timofeev E. N., Efimov N. V., Komyagin S. A. Podgotovka dannyh dlya sozdaniya cifrovogo proekta na uchastok remonta zheleznodorozhnogo puti s ispol'zovaniem SAU‑3D // Interekspo Geo-Sibir'. 2024. T. 1. S. 116– 121. DOI:https://doi.org/10.33764/2618-981X‑2024-1-116-121. EDN FQOUYJ
7. Baranov L. A., Ivanova N. D., Mihalevich I. F. Cifrovoy ispytatel'nyy stend analiza bezopasnosti ob'ektov kriticheskoy informacionnoy infrastruktury intellektual'nyh sistem vodnogo transporta // Nadezhnost'. 2025. T. 25, № 3. S. 50–59. DOI:https://doi.org/10.21683/1729-2646-2025-25-3-50-59. EDN QGPPKI
8. Kompleksnaya bezopasnost' ASU TP ob'ektov KII zheleznodorozhnogo transporta / P. A. Popov [i dr.] // Nadezhnost'. 2024. T. 24, № 4. S. 48–57. DOI:https://doi.org/10.21683/1729– 2646-2024-24-4-48-57. EDN YKHBOB
9. Baranov L. A., Safronov A. I., Sidorenko V. G. Razvitie intellektual'nyh sistem upravleniya elektricheskim transportom // Avtomatika, svyaz', informatika. 2025. № 10. S. 30–32. DOI:https://doi.org/10.62994/AT.2025.10.10.007. EDN BOSBZK
10. Baranov L. A., Kulagin M. A., Yanchenko G. O. Tehnicheskoe zrenie v intellektual'noy sisteme podderzhki prinyatiya resheniy mashinista special'noy putevoy mashiny // Nauka i tehnika transporta. 2025. № 3. S. 65– 69. EDN PZPESY
11. Kulagin M. A., Yanchenko G. O. Podhody k opredeleniyu prepyatstviy na zheleznodorozhnom puti s pomosch'yu neyronnyh setey // Intellektual'nye transportnye sistemy: materialy III Mezhdunarodnoy nauchno-prakticheskoy konferencii (Moskva, 30 maya 2024 goda). M.: Rossiyskiy universitet transporta (MIIT), 2024. S. 626–630. DOI:https://doi.org/10.30932/9785002446094-2024-626- 630. EDN OUMPCR
12. Nikul'chikov P. M. Istoriya, sostoyanie i perspektivy razvitiya sistem avtomaticheskogo upravleniya poezdami metropolitena // Avtomatika na transporte. 2016. T. 2, № 3. S. 456–473. EDN WYXBNL
13. Ozerov A. V., Bochkov A. V. Tekuschee sostoyanie i perspektivy mirovyh issledovaniy v oblasti intellektual'nogo zheleznodorozhnogo transporta // Intellektual'nyy transport. 2025. T. 9, № 2 (34). S. 43–66. EDN QFUBXM
14. Medvedeva E. V., Perevoschikova A. A. Issledovanie neyrosetevyh algoritmov raspoznavaniya ob'ektov zheleznodorozhnoy infrastruktury na videoizobrazheniyah // Komp'yuternaya optika. 2025. T. 49, № 3. S. 443–450. DOI:https://doi.org/10.18287/2412–6179-CO‑1563. EDN JUAWIT
15. Fedorov V. A., Ogorodnikova O. M. Segmentaciya ob'ektov tehnicheskim zreniem v avtomatizirovannyh sistemah upravleniya zheleznodorozhnym podvizhnym sostavom // Avtomatika na transporte. 2025. T. 11, № 3. S. 239–249. DOI:https://doi.org/10.20295/2412-9186-2025-11-03-239- 249. EDN IUNEIX
16. Deep Residual Learning for Image Recognition / K. He [et al.] // IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016 DOI:https://doi.org/10.1109/CVPR.2016.90
17. Tan M., Le Q. V. EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks // International Conference on Machine Learning (ICML). 2019. DOI:https://doi.org/10.48550/arXiv.1905.11946
18. Densely Connected Convolutional Networks / G. Huang [et al.] // IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2017. DOI:https://doi.org/10.1109/CVPR.2017.243
19. Ronneberger O., Fischer P., Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation // International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI). 2015. DOI:https://doi.org/10.1007/978-3-319-24574-4_28
20. UNet++: A Nested U-Net Architecture for Medical Image Segmentation / Z. Zhou [et al.] // International Workshop on Deep Learning in Medical Image Analysis (DLMIA/ MICCAI). 2018. DOI:https://doi.org/10.1007/978-3-030-00889-5_1
21. Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation / Chen L.-C. [et al.] // European Conference on Computer Vision (ECCV), 2018. DOI:https://doi.org/10.48550/arXiv.1802.02611
22. Multi-Attention-Network for Semantic Segmentation of Fine Resolution Remote Sensing Images / R. Li [et al.] // IEEE Transactions on Geoscience and Remote Sensing (TGRS). 2022. DOI:https://doi.org/10.1109/TGRS.2021.3093977
23. SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers / E. Xie [et al.] // Conference on Neural Information Processing Systems (NeurIPS). 2021. DOI:https://doi.org/10.48550/arXiv.2105.15203
24. Popov P. A., Ozerov A. V., Marshova A. S. Prospects of autonomous railway transport development, BRISS Transport. 2024 № 3 (3). Pp. 1–14. DOI:https://doi.org/10.46684/2024.3.4. EDN: https://elibrary.ru/HPYXEZ
