ForestFormer3D: A Unified Framework for End-to-End Segmentation of Forest LiDAR 3D Point Clouds

Binbin Xiang, Maciej Wielgosz, Stefano Puliti, Kamil Král, Martin Krůček, Azim Missarov, Rasmus Astrup · ICCV 2025 · 2025

DOI arXiv Code Project page

Status

To read

Topics

forest-segmentation, backbones

Feeds

P1 · ICCV 2027, P3 · Q1 journal

Dataset

FOR-instanceV2

What they did

Presents ForestFormer3D, an end-to-end framework for individual tree and semantic segmentation of forest LiDAR that combines ISA-guided query point selection, score-based block merging at inference and one-to-many association in training.

My note

Our closest application baseline: forest instance segmentation, instance-aware query selection, code and checkpoint instructions.

Code

Repository

SmartForest-no/ForestFormer3D

Framework

PyTorch

Licence

See the repository

My status

Not tried

BibTeX

@inproceedings{xiang2025-forestformer3d,
  title = {{ForestFormer3D: A Unified Framework for End-to-End Segmentation of Forest LiDAR 3D Point Clouds}},
  author = {Binbin Xiang and Maciej Wielgosz and Stefano Puliti and Kamil Král and Martin Krůček and Azim Missarov and Rasmus Astrup},
  booktitle = {ICCV 2025},
  year = {2025},
  doi = {10.1109/ICCV51701.2025.02292},
  eprint = {2506.16991},
  archivePrefix = {arXiv}
}