SynLiDAR is a synthetic LiDAR point cloud dataset that contains large-scale point-by-point annotated point clouds with accurate geometry and comprehensive semantic categories. The dataset collects 13 LiDAR point cloud sequences from virtual cities, suburban towns, communities, and ports, with about 20,000 scans (more than 19 billion points and 32 semantic categories).

Citation

@inproceedings{xiao2022transfer,
title={Transfer learning from synthetic to real LiDAR point cloud for semantic segmentation},
author={Xiao, Aoran and Huang, Jiaxing and Guan, Dayan and Zhan, Fangneng and Lu, Shijian},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
volume={36},
number={3},
pages={2795--2803},
year={2022}
}