Object-Level Data Augmentation for Deep Learning-Based Obstacle Detection in Railways
Source abstract
Abstract
This paper presents a novel method for generation of synthetic images of obstacles on and near rail tracks over long-range distances. The main goal is to augment the dataset for autonomous obstacle detection (OD) in railways, by inclusion of synthetic images that reflect the specific need for long-range OD in rail transport. The presented method includes a novel deep learning (DL)-based rail track detection that enables context- and scale-aware obstacle-level data augmentation. The augmented dataset is used for retraining of a state-of-the-art CNN for object detection. The evaluation results demonstrate significant improvement of detection of distant objects by augmentation of training dataset with synthetic images.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Infrastructure Maintenance and Monitoring · Advanced Neural Network Applications · Hand Gesture Recognition Systems
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Institute of Automation, University of Bremen, Otto-Hahn-Allee, NW1, 28359 Bremen, Germany
Funding
Shift2Rail Joint Undertaking under the European Union’s Horizon 2020 research and innovation program
References (30)
Show source-provided references
1. Analysis of the Safety Level of Obstacle Detection in Autonomous Railway Vehicles10.12700/APH.19.3.2022.3.15 ↗
2. (2022, February 28). ERRAC—Rail Strategic Research and Innovation Agenda, December 2020. Available online: https://errac.org/publications/rail-strategic-research-and-innovation-agenda-december-2020/.
3. Ristic-Durrant, D., Franke, M., and Michels, K. (2021). A Review of Vision-Based On-Board Obstacle Detection and Distance Estimation in Railways. Sensors, 21.10.3390/s21103452 ↗
4. Vision meets robotics: The KITTI dataset10.1177/0278364913491297 ↗
5. Zendel, O., Murschitz, M., Zeilinger, M., Steininger, D., Abbasi, S., and Beleznai, C. (2019, January 16–17). RailSem19: A Dataset for Semantic Rail Scene Understanding. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Long Beach, CA, USA.10.1109/CVPRW.2019.00161 ↗
6. Railway Traffic Object Detection Using Differential Feature Fusion Convolution Neural Network10.1109/TITS.2020.2969993 ↗
7. (2022, February 28). SMART—Smart Automation of Rail Transport. Available online: http://smart.masfak.ni.ac.rs/.
8. (2022, February 28). SMART2—Advanced Integrated Obstacle and Track Intrusion Detection System for Smart Automation of Rail Transport. Available online: https://smart2rail-project.net/.
9. SMART on-board multi-sensor obstacle detection system for improvement of rail transport safety
10. (2022, February 28). Hasty. Available online: https://hasty.ai/.
11. Martinez, R.B., Schiopu, I., Cornelis, B., and Munteanu, A. (2021). Real-Time Instance Segmentation of Traffic Videos for Embedded Deviced. Sensors, 21.10.3390/s21010275 ↗
12. Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollar, P., and Zitnick, C.L. (2014). Microsoft COCO: Common Objects in Context, in Computer Vision—ECCV 2014, Springer.10.1007/978-3-319-10602-1_48 ↗
13. Haseeb, M.A., Guan, J., Ristić-Durrant, D., and Gräser, A. (2018, January 1–5). DisNet: A novel method for distance estimation from monocular camera. Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems—IROS 2018, 10th Workshop on Planning, Perception and Navigation for Intelligent Vehicles (PPNIV), Madrid, Spain.
14. Zhou, X., Wang, D., and Krahenbühl, P. (2019). Objects as points. arXiv, Available online: https://arxiv.org/abs/1904.07850v2.
15. Franke, M., Gopinath, V., Reddy, C., Ristić-Durrant, D., and Michels, K. (2021, January 11–17). Bounding Box Dataset Augmentation for Long-range Object Distance Estimation. Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops 2021, Montreal, BC, Canada.10.1109/ICCVW54120.2021.00192 ↗
16. RailNet: A Segmentation Network for Railroad Detection10.1109/ACCESS.2019.2945633 ↗
17. Kaleli, F., and Akgul, Y.S. (2009, January 3–7). Vision-based railroad track extraction using dynamic programming. Proceedings of the 12th International IEEE Conference on Intelligent Transport Systems, St. Louis, MO, USA.10.1109/ITSC.2009.5309526 ↗
18. Context-guided ground truth sampling for multi-modality data augmentation in autonomous driving
19. Illarionova, S., Nesteruk, S., Shadrin, D., Ignatiev, V., Pukalchik, M., and Oseledets, I. (2021, January 11–17). Object-Based Augmentation for Building Semantic Segmentation: Ventura and Santa Rosa Case Study. Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), Montreal, BC, Canada.10.1109/ICCVW54120.2021.00191 ↗
20. Dwibedi, D., Misra, I., and Hebert, M. (2017, January 22–29). Cut, paste and learn: Surprisingly easy synthesis for instance detection. Proceedings of the IEEE International Conference on Computer Vision (ICCV), Venice, Italy.10.1109/ICCV.2017.146 ↗
21. Generative adversarial network: An overview of theory and applications
22. Rodriguez, L.A.F., Uribe, J.A., and Bonilla, J.F.V. (2012, January 12–14). Obstacle detection over rails using Hough transform. Proceedings of the XVII Symposium of Image, Signal Processing, and Artificial Vision (STSIVA), Medellin, Colombia.
23. Uribe, J.A., Fonseca, L., and Vargas, J.F. (2012, January 15–18). Video based system for railroad collision warning. Proceedings of the IEEE International Carnahan Conference on Security Technology (ICCST), Newton, MA, USA.10.1109/CCST.2012.6393573 ↗
24. Boussik, A., Ben-Messaoud, W., Niar, S., and Taleb-Ahmed, A. (2021, January 11–17). Railway obstacle detection using unsupervised learning: An exploratory study. Proceedings of the 32nd IEEE Intelligent Vehicles Symposium (IV), Nagoya, Japan.10.1109/IV48863.2021.9575825 ↗
25. Wu, H., Zheng, S., Zhang, J., and Huang, K. (2019, January 21–25). GP-GAN: Towards realistic high-resolution image blending. Proceedings of the 27th ACM International Conference on Multimedia, Nice, France.10.1145/3343031.3350944 ↗
26. Guo, B., Geng, G., Zhu, L., Shi, H., and Yu, Z. (2019). High-speed railway intruding object image generating with Generative Adversarial Networks. Sensors, 19.10.3390/s19143075 ↗
27. An Image Stitching Algorithm Based on Histogram Matching and SIFT Algorithm10.1142/S0218001417540064 ↗
28. Color transfer between images10.1109/38.946629 ↗
29. Szeliski, R. (2011). Computer Vision: Algorithms and Applications, Springer.10.1007/978-1-84882-935-0 ↗
30. (2022, February 28). UIC, Railway Transport Stock—Rolling Stock Construction Gauge. Available online: https://uic.org/com/enews/article/uic-publishes-first-edition-of-irs-50505-1-railway-transport-stock-rolling?var_recherche=505-1s.