artest08/OpenLane-V2-LiDAR

OpenLane-V2 LiDAR Extension

LiDAR point cloud data for OpenLane-V2, prepared for the multi-modal (camera + LiDAR) experiments in TopoBDA: Towards Bezier Deformable Attention for Road Topology Understanding, published in Neurocomputing.

• Paper: Neurocomputing (doi:10.1016/j.neucom.2025.132360) · arXiv:2412.18951
• Project page: artest08.github.io/TopoBDA.github.io
• Code: OpenLane-V2 fork, lidar_addition branch
• Tutorial: LiDAR integration notebook

This package extends an existing OpenLane-V2 download. It does not contain camera images or annotations: download the original OpenLane-V2 dataset separately, then merge this package into it.

Files

File	Contents	Size	md5
openlanev2_lidar.zip	LiDAR extension, extracts to OpenLane-V2/	45.8 GB	cd96d17bc6743b1732bc5634383c967a
readme.txt	Extraction and merge instructions	1.3 KB	08a755be32afc38a6fc35c5e28d8cb4e
md5sums.txt	Checksums for the files above

Download

With the Hugging Face CLI (pip install -U huggingface_hub):

hf download artest08/OpenLane-V2-LiDAR --repo-type dataset --local-dir ./openlanev2_lidar

Or with a direct link:

wget https://huggingface.co/datasets/artest08/OpenLane-V2-LiDAR/resolve/main/openlanev2_lidar.zip

Check the download, from the folder that holds it:

md5sum -c --ignore-missing md5sums.txt

Setup

The archive was created with zip -r -0 openlanev2_lidar.zip OpenLane-V2/, so it extracts to a folder named OpenLane-V2. Merge it into your existing OpenLane-V2 dataset without overwriting or removing existing files:

unzip openlanev2_lidar.zip -d <source_folder>
rsync -av <source_folder>/OpenLane-V2 <target_folder>

<target_folder> is the folder that contains your original OpenLane-V2 dataset. See the tutorial notebook for loading and visualizing the point clouds.

License

This extension is derived from OpenLane-V2 and is distributed under the same CC BY-NC-SA 4.0 license. Before using it, you must agree to the terms of use of the source datasets of OpenLane-V2: Argoverse 2 and nuScenes.

Citation

@article{kalfaoglu2026topobda,
  author  = {Muhammet Esat Kalfaoglu and Halil Ibrahim Ozturk and Ozsel Kilinc and Alptekin Temizel},
  title   = {{TopoBDA}: Towards {Bezier} deformable attention for road topology understanding},
  journal = {Neurocomputing},
  volume  = {670},
  pages   = {132360},
  year    = {2026},
  doi     = {10.1016/j.neucom.2025.132360},
  url     = {https://doi.org/10.1016/j.neucom.2025.132360}
}

@inproceedings{wang2023openlanev2,
  title={OpenLane-V2: A Topology Reasoning Benchmark for Unified 3D HD Mapping},
  author={Wang, Huijie and Li, Tianyu and Li, Yang and Chen, Li and Sima, Chonghao and Liu, Zhenbo and Wang, Bangjun and Jia, Peijin and Wang, Yuting and Jiang, Shengyin and Wen, Feng and Xu, Hang and Luo, Ping and Yan, Junchi and Zhang, Wei and Li, Hongyang},
  booktitle={NeurIPS},
  year={2023}
}