A Universal Railway Obstacle Detection System Based on Optical-Flow Guided Semi-Supervised Segmentation

Qiushi Guo, Bin Cao, Dehao Hao, Cheng Wang, Lijun Chen, Peng Yan

2025

Abstract

Detecting obstacles in railway scenarios is both crucial and challenging due to the wide range of obstacle categories and varying ambient conditions such as weather and light. Given the impossibility of encompassing all obstacle categories during the training stage, we address this out-of-distribution (OOD) issue with a semi-supervised segmentation approach guided by optical flow clues. We reformulate the task as a binary segmentation problem instead of the traditional object detection approach. To mitigate data shortages, we generate highly realistic synthetic images using Segment Anything (SAM) and YOLO, eliminating the need for manual annotation to produce abundant pixel-level annotations. Additionally, we leverage optical flow as prior knowledge to train the model effectively. Several experiments are conducted, demonstrating the feasibility and effectiveness of our approach.

Paper Citation

in Harvard Style

Guo Q., Cao B., Hao D., Wang C., Chen L. and Yan P. (2025). A Universal Railway Obstacle Detection System Based on Optical-Flow Guided Semi-Supervised Segmentation. In Proceedings of the 11th International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS; ISBN 978-989-758-745-0, SciTePress, pages 259-265. DOI: 10.5220/0013057600003941

in Bibtex Style

@conference{vehits25,
author={Qiushi Guo and Bin Cao and Dehao Hao and Cheng Wang and Lijun Chen and Peng Yan},
title={A Universal Railway Obstacle Detection System Based on Optical-Flow Guided Semi-Supervised Segmentation},
booktitle={Proceedings of the 11th International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS},
year={2025},
pages={259-265},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013057600003941},
isbn={978-989-758-745-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 11th International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS
TI - A Universal Railway Obstacle Detection System Based on Optical-Flow Guided Semi-Supervised Segmentation
SN - 978-989-758-745-0
AU - Guo Q.
AU - Cao B.
AU - Hao D.
AU - Wang C.
AU - Chen L.
AU - Yan P.
PY - 2025
SP - 259
EP - 265
DO - 10.5220/0013057600003941
PB - SciTePress