{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,29]],"date-time":"2026-05-29T18:07:19Z","timestamp":1780078039103,"version":"3.54.0"},"reference-count":51,"publisher":"Public Library of Science (PLoS)","issue":"5","license":[{"start":{"date-parts":[[2026,5,29]],"date-time":"2026-05-29T00:00:00Z","timestamp":1780012800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100024165","name":"Helmut-Schmidt-Universit\u00e4t\/Universit\u00e4t der Bundeswehr Hamburg","doi-asserted-by":"publisher","award":["Open-Access-Publication-Fund"],"award-info":[{"award-number":["Open-Access-Publication-Fund"]}],"id":[{"id":"10.13039\/100024165","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>Railways are among the most efficient modes of transportation, capable of moving large quantities of goods and passengers over long distances at relatively low cost. However, accidents frequently occur due to objects or individuals present on the tracks, as trains are unable to swerve and require long braking distances. While the localization of objects within the track bed is a well-explored topic, the reliable and high-performance classification of such obstacles across all relevant categories remains an unresolved challenge. This study proposes an innovative hybrid architecture that leverages the specific visual characteristics of track bed imagery, setting a new benchmark in this domain. The hybrid design effectively leverages the strengths of ResNet50 and Swin Transformer V2, allowing the model to capture both local and global features. In addition, an Efficient Attention Module is integrated to further emphasize the most relevant features for robust obstacle classification. Using stratified five-fold cross-validation on a dataset of 2,003 images across six classes (iron bar, boulder, person, branch, canister, and barrel), the model achieved an average balanced accuracy of 99.46%. The results have implications for accident prevention, improving operational efficiency, and modernizing railway safety systems, thereby enabling the future application of automatic railway surveillance systems to ultimately enhance operational security.<\/jats:p>","DOI":"10.1371\/journal.pone.0349562","type":"journal-article","created":{"date-parts":[[2026,5,29]],"date-time":"2026-05-29T17:29:37Z","timestamp":1780075777000},"page":"e0349562","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["Towards intelligent railway monitoring: A novel hybrid deep learning architecture for railway obstacle detection"],"prefix":"10.1371","volume":"21","author":[{"given":"Christopher","family":"Mai","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Luca","family":"Eisentraut","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Merlin","family":"Schadt","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2263-6408","authenticated-orcid":true,"given":"Ricardo","family":"Buettner","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,5,29]]},"reference":[{"key":"pone.0349562.ref001","doi-asserted-by":"crossref","first-page":"42","DOI":"10.1016\/j.tranpol.2023.07.007","article-title":"Efficiency of railway transport: A comparative analysis for 16 countries","volume":"141","author":"Y Niu","year":"2023","journal-title":"Transport Policy"},{"issue":"2","key":"pone.0349562.ref002","doi-asserted-by":"crossref","first-page":"346","DOI":"10.1109\/TITS.2015.2472580","article-title":"Text Mining the Contributors to Rail Accidents","volume":"17","author":"DE Brown","year":"2016","journal-title":"IEEE Trans Intell Transport Syst"},{"key":"pone.0349562.ref003","unstructured":"Eurostat. Railway safety statistics in the EU. 2024. https:\/\/ec.europa.eu\/eurostat\/statistics-explained\/index.php?title=Railway_safety_statistics_in_the_EU"},{"issue":"7","key":"pone.0349562.ref004","doi-asserted-by":"crossref","first-page":"6427","DOI":"10.1109\/TITS.2024.3412170","article-title":"Railway intrusion detection based on machine vision: A survey, challenges, and perspectives","volume":"25","author":"Y Wang","year":"2024","journal-title":"IEEE Trans Intell Transport Syst"},{"issue":"1","key":"pone.0349562.ref005","doi-asserted-by":"crossref","first-page":"101929","DOI":"10.1016\/j.jksuci.2024.101929","article-title":"RailSegVITNet: A lightweight VIT-based real-time track surface segmentation network for improving railroad safety","volume":"36","author":"Z Chen","year":"2024","journal-title":"J King Saud Univ Comput Inf Sci"},{"issue":"10","key":"pone.0349562.ref006","doi-asserted-by":"crossref","first-page":"17952","DOI":"10.1109\/TITS.2022.3156267","article-title":"A Stable Lightweight and Adaptive Feature Enhanced Convolution Neural Network for Efficient Railway Transit Object Detection","volume":"23","author":"T Ye","year":"2022","journal-title":"IEEE Trans Intell Transport Syst"},{"issue":"2","key":"pone.0349562.ref007","doi-asserted-by":"crossref","first-page":"4051","DOI":"10.3233\/JIFS-210349","article-title":"RSNet: Rail semantic segmentation network for extracting aerial railroad images","volume":"41","author":"RS Rampriya","year":"2021","journal-title":"IFS"},{"issue":"11","key":"pone.0349562.ref008","doi-asserted-by":"crossref","first-page":"676","DOI":"10.3390\/drones8110676","article-title":"A Fast Obstacle Detection Algorithm Based on 3D LiDAR and Multiple Depth Cameras for Unmanned Ground Vehicles","volume":"8","author":"F Pang","year":"2024","journal-title":"Drones"},{"issue":"5","key":"pone.0349562.ref009","doi-asserted-by":"crossref","DOI":"10.3390\/electronics12051256","article-title":"SDRC-YOLO: A Novel Foreign Object Intrusion Detection Algorithm in Railway Scenarios","volume":"12","author":"C Meng","year":"2023","journal-title":"Electronics"},{"issue":"14","key":"pone.0349562.ref010","doi-asserted-by":"crossref","first-page":"4483","DOI":"10.3390\/s24144483","article-title":"Research on the Method of Foreign Object Detection for Railway Tracks Based on Deep Learning","volume":"24","author":"S Ning","year":"2024","journal-title":"Sensors (Basel)"},{"key":"pone.0349562.ref011","unstructured":"Wang T, Zhang Z, Yang F, Tsui KL. Intelligent Railway Foreign Object Detection: A Semi-supervised Convolutional Autoencoder Based Method. arXiv preprint. 2021;2108.02421."},{"issue":"1","key":"pone.0349562.ref012","doi-asserted-by":"crossref","first-page":"1315","DOI":"10.1038\/s41597-024-03952-3","article-title":"An unmanned aerial vehicle captured dataset for railroad segmentation and obstacle detection","volume":"11","author":"R R S","year":"2024","journal-title":"Sci Data"},{"issue":"27","key":"pone.0349562.ref013","doi-asserted-by":"crossref","first-page":"145","DOI":"10.1016\/j.ifacol.2019.12.747","article-title":"Monitoring the condition of the protective fence above the railway track","volume":"52","author":"R Hajovsky","year":"2019","journal-title":"IFAC-PapersOnLine"},{"issue":"1","key":"pone.0349562.ref014","doi-asserted-by":"crossref","first-page":"137","DOI":"10.1016\/j.retrec.2012.12.003","article-title":"The economics of railway safety","volume":"43","author":"AW Evans","year":"2013","journal-title":"Res Transp Econ"},{"key":"pone.0349562.ref015","doi-asserted-by":"crossref","first-page":"104242","DOI":"10.1016\/j.autcon.2022.104242","article-title":"Developments, challenges, and perspectives of railway inspection robots","volume":"138","author":"G Jing","year":"2022","journal-title":"Autom Constr"},{"key":"pone.0349562.ref016","doi-asserted-by":"crossref","first-page":"85761","DOI":"10.1109\/ACCESS.2024.3415035","article-title":"A dense attention railway foreign object detection algorithm based on mask R-CNN","volume":"12","author":"S Gao","year":"2024","journal-title":"IEEE Access"},{"key":"pone.0349562.ref017","doi-asserted-by":"crossref","first-page":"143772","DOI":"10.1109\/ACCESS.2019.2945633","article-title":"RailNet: A segmentation network for railroad detection","volume":"7","author":"Y Wang","year":"2019","journal-title":"IEEE Access"},{"key":"pone.0349562.ref018","doi-asserted-by":"crossref","first-page":"109241","DOI":"10.1016\/j.measurement.2021.109241","article-title":"Obstacle detection of rail transit based on deep learning","volume":"176","author":"D He","year":"2021","journal-title":"Measurement"},{"issue":"5","key":"pone.0349562.ref019","doi-asserted-by":"crossref","DOI":"10.1007\/s11554-024-01548-z","article-title":"A real-time foreign object detection method based on deep learning in complex open railway environments","volume":"21","author":"B Zhang","year":"2024","journal-title":"J Real-Time Image Proc"},{"key":"pone.0349562.ref020","doi-asserted-by":"crossref","unstructured":"Xu Y, Gao C, Yuan L, Tang S, Wei G. Real-time Obstacle Detection Over Rails Using Deep Convolutional Neural Network. In: 2019 IEEE Intelligent Transportation Systems Conference (ITSC). 2019. p. 1007\u201312.","DOI":"10.1109\/ITSC.2019.8917091"},{"key":"pone.0349562.ref021","doi-asserted-by":"crossref","first-page":"18498","DOI":"10.1109\/ACCESS.2025.3533304","article-title":"Railway Foreign Object Intrusion Detection Using UAV Images and YOLO-UAT","volume":"13","author":"Y Yang","year":"2025","journal-title":"IEEE Access"},{"key":"pone.0349562.ref022","first-page":"19","article-title":"Detection of Foreign Objects Around the Railway Line with YOLOv8","author":"M Sevi","year":"2023","journal-title":"JCS"},{"key":"pone.0349562.ref023","doi-asserted-by":"crossref","unstructured":"Qi Z, Ma D, Xu J, Xiang A, Qu H. Improved YOLOv5 Based on Attention Mechanism and FasterNet for Foreign Object Detection on Railway and Airway tracks. 2024. p. 1\u20136.","DOI":"10.1109\/ASIANComNet63184.2024.10811086"},{"issue":"12","key":"pone.0349562.ref024","doi-asserted-by":"crossref","first-page":"13442","DOI":"10.1109\/JSEN.2021.3066714","article-title":"A Camera and LiDAR Data Fusion Method for Railway Object Detection","volume":"21","author":"W Zhangyu","year":"2021","journal-title":"IEEE Sensors J"},{"key":"pone.0349562.ref025","doi-asserted-by":"crossref","unstructured":"Ding X, Cai X, Zhang Z, Liu W, Song W. Railway Foreign Object Intrusion Detection based on Deep Learning. In: 2022 International Conference on Computer Engineering and Artificial Intelligence (ICCEAI). 2022. p. 735\u20139.","DOI":"10.1109\/ICCEAI55464.2022.00155"},{"key":"pone.0349562.ref026","first-page":"3749635","article-title":"Foreign Object Detection in Railway Images Based on an Efficient Two-Stage Convolutional Neural Network","volume":"2022","author":"W Chen","year":"2022","journal-title":"Comput Intell Neurosci"},{"issue":"12","key":"pone.0349562.ref027","doi-asserted-by":"crossref","first-page":"1055","DOI":"10.3390\/mi11121055","article-title":"Object Detection at Level Crossing Using Deep Learning","volume":"11","author":"MAB Fayyaz","year":"2020","journal-title":"Micromachines (Basel)"},{"issue":"15","key":"pone.0349562.ref028","doi-asserted-by":"crossref","first-page":"21083","DOI":"10.1007\/s11042-022-12059-z","article-title":"An intelligent railway surveillance framework based on recognition of object and railway track using deep learning","volume":"81","author":"R Kapoor","year":"2022","journal-title":"Multimed Tools Appl"},{"key":"pone.0349562.ref029","doi-asserted-by":"crossref","unstructured":"Deng J, Dong W, Socher R, Li LJ, Li K, Fei-Fei L. ImageNet: A Large-Scale Hierarchical Image Database. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition. 2009. p. 248\u201355. doi:10.1109\/CVPR.2009.5206848","DOI":"10.1109\/CVPR.2009.5206848"},{"issue":"8","key":"pone.0349562.ref030","doi-asserted-by":"crossref","first-page":"5200","DOI":"10.1109\/TNNLS.2021.3119238","article-title":"Spiking Deep Residual Networks","volume":"34","author":"Y Hu","year":"2023","journal-title":"IEEE Trans Neural Netw Learn Syst"},{"key":"pone.0349562.ref031","doi-asserted-by":"crossref","unstructured":"He K, Zhang X, Ren S, Sun J. Deep Residual Learning for Image Recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016. p. 770\u20138.","DOI":"10.1109\/CVPR.2016.90"},{"key":"pone.0349562.ref032","doi-asserted-by":"crossref","first-page":"155","DOI":"10.1016\/j.neunet.2022.01.012","article-title":"Low-degree term first in ResNet, its variants and the whole neural network family","volume":"148","author":"T Sun","year":"2022","journal-title":"Neural Netw"},{"key":"pone.0349562.ref033","article-title":"Improved deep learning image classification algorithm based on Swin Transformer V2","volume":"9","author":"J Wei","year":"2023","journal-title":"PeerJ Comput Sci"},{"key":"pone.0349562.ref034","article-title":"Swin Transformer: Hierarchical Vision Transformer using Shifted Windows","author":"Z Liu","year":"2021","journal-title":"CoRR"},{"key":"pone.0349562.ref035","first-page":"448","article-title":"Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift.","volume-title":"Proceedings of the 32nd International Conference on Machine Learning. vol. 37 of Proceedings of Machine Learning Research","author":"S Ioffe","year":"2015"},{"key":"pone.0349562.ref036","doi-asserted-by":"crossref","first-page":"108","DOI":"10.1016\/j.patrec.2019.11.015","article-title":"Selective feature connection mechanism: concatenating multi-layer CNN features with a feature selector","volume":"129","author":"C Du","year":"2020","journal-title":"Pattern Recognit Lett"},{"issue":"1","key":"pone.0349562.ref037","doi-asserted-by":"crossref","first-page":"31941","DOI":"10.1038\/s41598-024-83430-3","article-title":"Feature optimization-guided high-precision and real-time metal surface defect detection network","volume":"14","author":"S Chan","year":"2024","journal-title":"Sci Rep"},{"key":"pone.0349562.ref038","doi-asserted-by":"crossref","unstructured":"Mascarenhas S, l Agarwal M. A comparison between VGG16, VGG19 and ResNet50 architecture frameworks for Image Classification. 2021 International Conference on Disruptive Technologies for Multi-Disciplinary Research and Applications (CENTCON). 2021;1:96\u20139.","DOI":"10.1109\/CENTCON52345.2021.9687944"},{"key":"pone.0349562.ref039","doi-asserted-by":"crossref","unstructured":"Wang Q, Wu B, Zhu P, Li P, Zuo W, Hu Q. ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks. In: Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE; 2020. p. 11531\u201339.","DOI":"10.1109\/CVPR42600.2020.01155"},{"key":"pone.0349562.ref040","first-page":"2825","article-title":"Scikit-learn: Machine Learning in Python","volume":"12","author":"F Pedregosa","year":"2011","journal-title":"J Mach Learn Res"},{"issue":"12","key":"pone.0349562.ref041","doi-asserted-by":"crossref","DOI":"10.1115\/1.4044524","article-title":"A Comparative Evaluation of Supervised Machine Learning Classification Techniques for Engineering Design Applications","volume":"141","author":"C Sharpe","year":"2019","journal-title":"J Mech Design"},{"key":"pone.0349562.ref042","doi-asserted-by":"crossref","first-page":"43201","DOI":"10.1109\/ACCESS.2025.3549271","article-title":"Improvement of Deep Learning Models Using Retinal Filter: A Systematic Evaluation of the Effect of Gaussian Filtering With a Focus on Industrial Inspection Data","volume":"13","author":"R Buettner","year":"2025","journal-title":"IEEE Access"},{"issue":"3","key":"pone.0349562.ref043","doi-asserted-by":"crossref","first-page":"533","DOI":"10.3390\/rs16030533","article-title":"Selecting and Interpreting Multiclass Loss and Accuracy Assessment Metrics for Classifications with Class Imbalance: Guidance and Best Practices","volume":"16","author":"S Farhadpour","year":"2024","journal-title":"Remote Sens"},{"key":"pone.0349562.ref044","first-page":"2825","article-title":"Scikit-learn: Machine Learning in Python","volume":"12","author":"F Pedregosa","year":"2011","journal-title":"J Mach Learn Res"},{"issue":"4","key":"pone.0349562.ref045","doi-asserted-by":"crossref","first-page":"427","DOI":"10.1016\/j.ipm.2009.03.002","article-title":"A systematic analysis of performance measures for classification tasks","volume":"45","author":"M Sokolova","year":"2009","journal-title":"Inf Process Manag"},{"issue":"5","key":"pone.0349562.ref046","doi-asserted-by":"crossref","DOI":"10.3390\/medicina57050503","article-title":"Foundational statistical principles in medical research: sensitivity, specificity, positive predictive value, and negative predictive value","volume":"57","author":"TF Monaghan","year":"2021","journal-title":"Medicina"},{"issue":"21","key":"pone.0349562.ref047","doi-asserted-by":"crossref","DOI":"10.3390\/app14219863","article-title":"Performance metrics for multilabel emotion classification: comparing micro, macro, and weighted f1-scores","volume":"14","author":"MC Hinojosa Lee","year":"2024","journal-title":"Appl Sci"},{"issue":"1","key":"pone.0349562.ref048","doi-asserted-by":"crossref","first-page":"37","DOI":"10.1177\/001316446002000104","article-title":"A Coefficient of Agreement for Nominal Scales","volume":"20","author":"J Cohen","year":"1960","journal-title":"Educ Psychol Meas"},{"issue":"1","key":"pone.0349562.ref049","doi-asserted-by":"crossref","first-page":"151","DOI":"10.1111\/tgis.12204","article-title":"Optimal Stopping Location of a High Speed Train using GIS and Multicriteria Decision\u2010making","volume":"21","author":"R Sa\u00f1udo","year":"2017","journal-title":"Trans GIS"},{"key":"pone.0349562.ref050","unstructured":"Auditors E. Rail freight transport in the EU: still not on the right track. Special Report. 2016."},{"key":"pone.0349562.ref051","doi-asserted-by":"crossref","first-page":"28761","DOI":"10.1109\/ACCESS.2024.3365274","article-title":"A Novel Hybrid Deep Learning Architecture for Dynamic Hand Gesture Recognition","volume":"12","author":"DRT Hax","year":"2024","journal-title":"IEEE Access"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0349562","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,29]],"date-time":"2026-05-29T17:29:47Z","timestamp":1780075787000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0349562"}},"subtitle":[],"editor":[{"given":"Armin","family":"Moghimi","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,5,29]]},"references-count":51,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2026,5,29]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0349562","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,29]]}}}