Research and Analysis of Railway Obstacle Detection Technology Based on Deep Learning
DOI:
https://doi.org/10.61173/p3eb3n60Keywords:
Deep Learning, Obstacle Detection, Convo-lutional Neural Networks, Rail TransitAbstract
The advancement of intelligent detection technologies has made highly reliable obstacle identification a critical factor in enhancing the operational safety of railways. Nevertheless, traditional detection methods (such as railside infrared counters and vehicle-mounted millimeterwave radars) are limited by inherent defects such as strong environmental sensitivity, high detection rate of small targets and no classification ability, making it difficult to meet the safety needs in complex scenarios. In recent years, deep learning has promoted the transition of railway obstacle detection technology to intelligence with its multimodal feature autonomous extraction and dynamic scene modeling capabilities. This review offers a thorough analysis of the historical development and advancements in railway obstacle detection methods, highlighting the transformative advances brought about by deep learningbased models, comprehensively compares the application characteristics and scenarios of various technologies, and discusses the application and evolution of deep learning in the two core tasks of target recognition and semantic segmentation, covering representative models such as convolutional neural network and Transformer. This article also uses the analysis of public data sets to help the new generation of railway obstacle detection technology to provide a theoretical basis and also facilitate researchers to quickly carry out experimental verification.
References
[1] Zhang Q, Yan F, Song W, Wang R, Li G. Automatic Obstacle Detection Method for the Train Based on Deep Learning. Sustainability. 2023, 15, 1184.
[2] Bradski G. The opencv library. Dr. Dobb’s Journal: Software Tools for the Professional Programmer. 2000, 25(11): 120-123.
[3] Ren S, He K, Girshick R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE transactions on pattern analysis and machine intelligence. 2016, 39(6): 1137-1149.
[4] Redmon J, Divvala S, Girshick R, et al. You only look once: Unified, real-time object detection. Proceedings of the IEEE conference on computer vision and pattern recognition. 2016, 779-788.
[5] Redmon J, Farhadi A. Yolov3: An incremental improvement. arxiv preprint arxiv:1804.02767, 2018.
[6] Cui Y, Chen R, Chu W, et al. Deep learning for image and point cloud fusion in autonomous driving: A review. IEEE Transactions on Intelligent Transportation Systems. 2021, 23(2): 722-739.
[7] Bai X, Hu Z, Zhu X, et al. Transfusion: Robust lidar-camera fusion for 3d object detection with transformers. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022, 1090-1099.
[8] Wang C Y, Yeh I H, Mark Liao H Y. Yolov9: Learning what you want to learn using programmable gradient information. European conference on computer vision. Cham: Springer Nature Switzerland. 2024, 1-21.
[9] Ke L, Danelljan M, Li X, et al. Mask Transfiner for High- Quality Instance Segmentation. arXiv, 2021(2021).
[10] Cao C, Wang B, Zhang W, et al. An improved faster R-CNN for small object detection. Ieee Access. 2019, 7: 106838-106846.
[11] Huangfu JY, Meng Q, Meng LC, Xie YP. YOLOv5 Traffic Object Detection Based on GhostNet and Attention Mechanism. Computer Systems and Applications, 2023, 32(4): 149-160(in Chinese).
[12] Zhao Z, Kang J, Wu B, et al. AE-Net: a high accuracy and efficient network for railway obstacle detection based on convolution and transformer. IEEE Transactions on Instrumentation and Measurement. 2024, 73: 1-14.
[13] Li R, He C, Li S, et al. DynaMask: dynamic mask selection for instance segmentation. roceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2023, 11279-11288.
[14] Wang W, Dai J, Chen Z, et al. Internimage: Exploring largescale vision foundation models with deformable convolutions. Proceedings of the IEEE/CVF conference on computer vision Dean&Francis ISSN 2959-6157 and pattern recognition. 2023, 14408-14419.
[15] Wei Z, Liu L, Hu Q X, et al. Methods, devices, equipment and media for detecting and positioning of railway unmanned driving obstacles: CN202510596776.2. CN120107933A [2025- 09-26].
[16] Tagiew R, Klasek P, Tilly R, et al. Osdar23: Open sensor data for rail 2023. 2023 8th International Conference on Robotics and Automation Engineering (ICRAE). IEEE, 2023, 270-276.
[17] Diaz X, D’Amico G, Dominguez-Sanchez R, et al. Towards Railway Domain Adaptation for LiDAR-based 3D Detection: Road-to-Rail and Sim-to-Real via SynDRA-BBox. arxiv preprint arxiv:2507, 16413, 2025.
[18] Li X, Peng X. Rail detection: An efficient row-based network and a new benchmark. Proceedings of the 30th ACM International Conference on Multimedia. 2022, 6455-6463.
[19] Zhou D W, Zhang Y, Wang Y, et al. Learning without forgetting for vision-language models. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2025.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 by the authors.
This work is licensed under a Creative Commons Attribution 4.0 International License.
