Inicio  /  Infrastructures  /  Vol: 9 Par: 4 (2024)  /  Artículo
ARTÍCULO
TITULO

Multi-Context Point Cloud Dataset and Machine Learning for Railway Semantic Segmentation

Abderrazzaq Kharroubi    
Zouhair Ballouch    
Rafika Hajji    
Anass Yarroudh and Roland Billen    

Resumen

Railway scene understanding is crucial for various applications, including autonomous trains, digital twining, and infrastructure change monitoring. However, the development of the latter is constrained by the lack of annotated datasets and limitations of existing algorithms. To address this challenge, we present Rail3D, the first comprehensive dataset for semantic segmentation in railway environments with a comparative analysis. Rail3D encompasses three distinct railway contexts from Hungary, France, and Belgium, capturing a wide range of railway assets and conditions. With over 288 million annotated points, Rail3D surpasses existing datasets in size and diversity, enabling the training of generalizable machine learning models. We conducted a generic classification with nine universal classes (Ground, Vegetation, Rail, Poles, Wires, Signals, Fence, Installation, and Building) and evaluated the performance of three state-of-the-art models: KPConv (Kernel Point Convolution), LightGBM, and Random Forest. The best performing model, a fine-tuned KPConv, achieved a mean Intersection over Union (mIoU) of 86%. While the LightGBM-based method achieved a mIoU of 71%, outperforming Random Forest. This study will benefit infrastructure experts and railway researchers by providing a comprehensive dataset and benchmarks for 3D semantic segmentation. The data and code are publicly available for France and Hungary, with continuous updates based on user feedback.

 Artículos similares

       
 
Massimiliano Pepe, Domenica Costantino and Vincenzo Saverio Alfio    
The aim of the paper is to identify a suitable method for assessing the deformation of structures (buildings, bridges, walls, etc.) by means of topographic measurements of significant targets positioned on the infrastructure under consideration. In parti... ver más
Revista: Infrastructures

 
Mohammed Y. Fattah, Qutaiba G. Majeed and Hasan H. Joni    
The theoretical and practical studies of the cyclic loads resulting from the movement and passage of trains on the unsaturated subgrade to determine the effect of the degree of saturation and moisture content on the foundations and infrastructure of the ... ver más
Revista: Infrastructures

 
Felix Lampe, Maren Maus, Lea Elfert and Nils Nießen    
This paper presents a model that establishes a relationship between the financial resources allocated to maintain and replace railway station infrastructure assets and the resulting infrastructure quality. Until now, there has been no known relationship ... ver más
Revista: Applied Sciences

 
Ze Yu, Chuxin Wang, Yuanyuan Zhao, Zhiyuan Hu and Yuanjie Tang    
The linear scheduling method (LSM) for optimization in linear projects has been the focus of numerous academic studies over the years. However, research on incorporating reverse construction activities and other practical scenarios, such as flexible acti... ver más
Revista: Applied Sciences

 
Stefan Marschnig, Markus Loidolt, Dieter Knabl, Alwine Steinecker and Reinhard Popp    
Managing head checks is a crucial task for an infrastructure manager as in case of deep cracks, rails can break and, thus, accidents might be the consequence. Many infrastructure managers use vehicle-mounted eddy-current testing for detecting cracks. Thi... ver más
Revista: Infrastructures