RailCloud-HdF: A Large-Scale Point Cloud Dataset for Railway Scene Semantic Segmentation

Résumé fourni par la source

RailCloud-HdF is a large-scale LiDAR point-cloud dataset for railway scene semantic segmentation. It was introduced in the VISAPP 2024 paper “RailCloud-HdF: A Large-Scale Point Cloud Dataset for Railway Scene Semantic Segmentation” by Mahdi Abid, Mathis Teixeira, Ankur Mahtani, and Thomas Laurent. The dataset contains railway LiDAR point clouds stored as compressed `.laz` files and organized into eight acquisition folders corresponding to railway lines in the Hauts-de-France region. Each acquisition folder contains a `cloud/` directory with the point-cloud files and an `info/` directory with scan-level metadata files. The dataset covers approximately 267.52 km of railway scenes and contains 5,353 point-cloud tiles, for a total of about 8.06 billion points. This release also includes the original train/validation/test split files, scan-level and acquisition-level manifests, and SHA-256 checksums for verifying the uploaded archives. Companion repository containing release scripts, metadata, manifests, and split utilities:https://github.com/mahdiabid91/RailCloud-HdF If you use RailCloud-HdF in your research, please cite the associated paper: Mahdi Abid, Mathis Teixeira, Ankur Mahtani, and Thomas Laurent. “RailCloud-HdF: A Large-Scale Point Cloud Dataset for Railway Scene Semantic Segmentation.” In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP, pages 159–170, 2024. DOI: 10.5220/0012394800003660. The Zenodo DOI identifies this archived dataset release and can be used to reference the exact dataset version.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.