Published 2025 | Version v2
Dataset Open

SemanticRail3D: A 3D Point Cloud dataset with semantic annotations of railway environments

Description

This repository contains the SemanticRail3D dataset, a high-quality 3D point cloud dataset for railway infrastructure segmentation and classification. It consists of 438 point clouds, each covering 200 meters of track, totaling 2.8 billion points labeled into 11 semantic classes. The dataset is structured for machine learning applications, with train, validation, and test splits.

 

More info in our github repository.

Files

SemanticRail3D_v2.zip

Files (7.2 GB)

Name Size Download all
md5:ec5baf7458c9b4357592c2b38f81c686
7.2 GB Preview Download

Additional details

Related works

Is derived from
Journal article: 10.3390/rs13122332 (DOI)
Is referenced by
Journal article: 10.1016/j.autcon.2023.104854 (DOI)

Funding

Agencia Estatal de Investigación
Mantenimiento predictivo-cognitivo para la gestión integrada del ferrocarril (4RAIL) CPP2021-008374
Agencia Estatal de Investigación
RYC2021‐033560‐I

Software