Mangrove3D is a terrestrial laser scanning (TLS) dataset for 3D point cloud semantic segmentation, collected in spring 2024 on Babeldaob Island, Palau (7°31′49″N, 134°33′53″E). The study sites consist of coastal Rhizophora mangrove stands characterized by dense prop-root systems and multilayered canopies. Data were acquired using a Canopy Biomass LiDAR v2.0 (CBL) system based on a SICK LMS151 scanner (905 nm), mounted on a rotating stage and inverted on a surface elevation table (SET) arm to maximize near-ground and root sampling. Each scan covers a 360° × 270° field of view at 0.25° angular resolution and is completed in approximately 33 s. At each benchmark site, eight scans were collected at 45° azimuthal intervals; scans exhibiting obvious acquisition errors were excluded. The dataset contains 39 scans from 7 benchmark sites, comprising approximately 31.3 million points (0.5–0.9 M points per scan). Points are annotated into six semantic classes: Ground & Water, Stem, Canopy, Root, Object, and Void. For reproducible benchmarking, we recommend using 30 scans (sites #1–5) for training and validation and 9 scans (sites #6–7) for testing. Mangrove3D is among the first open TLS datasets specifically targeting mangrove ecosystems. It provides high-density 3D geometry suitable for semantic segmentation, structural analysis, biomass estimation, and coastal blue-carbon research. Dataset Highlights High-resolution TLS dataset dedicated to body-level semantic segmentation of mangrove forests, including explicit annotation of prop-root structures. One of the first open-access 3D point cloud datasets capturing coastal Rhizophora mangrove stands. Points annotated into 5 classes: Ground&Water, stem, canopy, root, and object. Spherical projection maps with corresponding ground-truth semantic segmentation masks provided. We also have a project web page for this dataset: https://fz-rit.github.io/through-the-lidars-eye/ If you use this dataset in your research, please consider citing our work: @dataset{zhang_2026_18510441, author = {Zhang, Fei and Chancia, Robert and Clapp, Josie and Hassanzadeh, Amirhossein and Dera, Dimah and MacKenzie, Richard and van Aardt, Jan}, title = {Mangrove3D: Terrestrial Laser Scanning Dataset for Coastal Mangrove Forests }, month = feb, year = 2026, publisher = {Zenodo}, doi = {10.5281/zenodo.18510441}, url = {https://doi.org/10.5281/zenodo.18510441},} @misc{zhang2025perspectivelidarfeatureenricheduncertaintyaware, title={Through the Perspective of LiDAR: A Feature-Enriched and Uncertainty-Aware Annotation Pipeline for Terrestrial Point Cloud Segmentation}, author={Fei Zhang and Rob Chancia and Josie Clapp and Amirhossein Hassanzadeh and Dimah Dera and Richard MacKenzie and Jan van Aardt}, year={2025}, eprint={2510.06582}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2510.06582}, }