Abstract
Potential safety hazards along the track primarily involve identifying lightweight objects within a 500-m radius on both sides. The existing manual inspection procedure is time-consuming and poses risks, particularly in remote regions. Unmanned aerial vehicles (UAV) offer a viable solution, efficiently conducting railway inspections with their wide field of view and flexibility. Image segmentation yields crucial, precise features essential for drone inspections. However, conventional techniques encounter challenges with drone imagery due to limited and non-diverse samples, hindering their ability to generalize. Additionally, these samples do not encompass all potential objects, necessitating frequent model updates—an arduous task. Accurately gauging the distance between the target and the railway track remains a challenge for precise risk assessment. To address these concerns, we introduce a method for segmenting potential hazards in the railway vicinity, utilizing the Fast Segment Anything Model (Fast SAM). This approach unfolds in three stages. First, data captured by drones undergoes image segmentation using the Fast SAM model, producing object masks. Subsequently, through various image processing techniques, we extract center points of hazards and endpoints of railway tracks from these segmented masks. This data facilitates distance computation between hazards and the railway using appropriate mathematical formulas. Lastly, hazard levels are determined based on predefined criteria. This comprehensive method augments hazard detection, inspection efficiency, and the accuracy of risk assessment within the railway vicinity.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Xu, Y., Liu, J.: High-speed train fault detection with unsupervised causality-based feature extraction methods. Adv. Eng. Inform. 49, 101312 (2021)
General Office of the State Council, 2021. Accessed 1 April 2022 (in Chinese)
Wu, Y., Qin, Y., Qian, Y., Guo, F.: Automatic detection of arbitrarily oriented fastener defect in high-speed railway. Autom. Constr. 131 (2021)
Wu, Y., Qin, Y., Wang, Z., Jia, L.: A UAV-based visual inspection method for rail surface defects. Appl. Sci.-Basel 8(7) (2018)
Yi, W., Sutrisna, M., Wang, H.: Unmanned aerial vehicle based low carbon monitoring planning. Adv. Eng. Inf. 48, 101277 (2021)
Zhang, J., Zhao, J., Zhang, R., Lv, X., Nie, j.: Overview of deep learning image instance segmentation methods. J. Small Microcomput. Syst. 42(1), 161–171 (2021)
He, K., Gkioxari, G., Dollár, P., Girshick, R.: Mask r-cnn. arXiv 2017. user-6073b1344c775e0497f43bf9 (2020)
Huang, Z., Huang, L., Gong, Y., Huang, C., Wang, X.: Mask Scoring R-Cnn. Computer Vision and Pattern Recognition abs/1903.00241, 6409–6418 (2019)
Bolya, D., Zhou, C., Xiao, F., Lee, Y.: YOLACT++ better real-time instance segmentation. IEEE Trans. Pattern Anal. Mach. Intell. 44(2), 9156–9165 (2022)
Bolya, D., Zhou, C., Xiao, F., Lee, Y.: YOLACT: Real-time instance segmentation. CoRR abs/1904.02689 (2019)
Liu, S., Jia, J., Fidler, S., Urtasun, R.: Sgn: sequential grouping networks for instance segmentation. IEEE International Conference on Computer Vision 2017(1), 3516–3524 (2017)
Gao, N., et al.: SSAP: single-shot instance segmentation with affinity pyramid. IEEE International Conference on Computer Vision 2019(1), 642–651 (2019)
Peng, S., Jiang, W., Pi, H., Bao, H., Zhou, X.: Deep snake for real-time instance segmentation. CVPR 2020, 8530–8539 (2020)
Wang, X., Kong, T., Shen, C., Jiang, Y., Li, L.: SOLO: segmenting objects by locations. In: European Conference on Computer Vision, pp. 649–665 (2020)
Wang, X., Zhang, R., Kong, T., Li, L., Shen, C.: SOLOv2—Dynamic and fast instance segmentation. Conference on Neural Information Processing Systems 33, 17721–17732 (2020)
Kirillov, A., Eric, M., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Alexander, C., Lo, W., Dollár, P., Girshick, R.: Segment Anything. CoRR, abs/2304.02643 (2023)
He, K., Chen, X., Xie, S., Li, Y., Dollár, P., Girshick, R.: Masked autoencoders are scalable vision learners. Computer Vision and Pattern Recognition 2022(1), 15979–15988 (2022)
Zhao, X., Ding, W., An, Y., Du, Y., Yu, T., Li, M., Tang, M., Wang, J.: Fast Segment Anything. CoRR, abs/2306.12156 (2023)
Zhang, C., Han, D., Qiao, Y., Kim, J., Bae, S., Lee, S., Hong, C.: Faster segment anything: towards lightweight SAM for mobile applications. CoRR. abs/2306.14289 (2023)
High-Speed Railway Safety Protection and Management Methods (2020) (in Chinese)
Wu, Y., Meng, F., Qin, Y., Qian, Y., Xu, F., Jia, L.: UAV imagery based potential safety hazard evaluation for high-speed railroad using Real-time instance segmentation. Adv. Eng. Inf. 55 (2023)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2024 Beijing Paike Culture Commu. Co., Ltd.
About this paper
Cite this paper
Li, S., Yu, C., Chang, L., Zhao, X. (2024). Railway Surrounding Environment Hazard Detection Based on Fast SAM. In: Yang, J., Yao, D., Jia, L., Qin, Y., Liu, Z., Diao, L. (eds) Proceedings of the 6th International Conference on Electrical Engineering and Information Technologies for Rail Transportation (EITRT) 2023. EITRT 2023. Lecture Notes in Electrical Engineering, vol 1136. Springer, Singapore. https://doi.org/10.1007/978-981-99-9315-4_63
Download citation
DOI: https://doi.org/10.1007/978-981-99-9315-4_63
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-99-9314-7
Online ISBN: 978-981-99-9315-4
eBook Packages: EngineeringEngineering (R0)