Sensor-agnostic 3D structure mapping pipeline: from UAS-derived 2D orthophotos and LiDAR point clouds to volumetric training data for 3D U-Net modelling
We present a reusable, sensor-agnostic data pipeline that integrates diverse passive (e.g., multispectral orthophotos) and active (e.g., LiDAR point clouds) remote sensing datasets collected from uncrewed aerial systems (UAS). This pipeline enables the generation of volumetric vegetation structure labels by combining 2D UAS-derived imagery with 3D structural information.
Authors
- Teja Kattenborn (ORCID: https://orcid.org/0000-0001-7381-3828)
- Laura N. Sotomayor (ORCID: https://orcid.org/0000-0001-9791-2193)
- Darren Turner (ORCID: https://orcid.org/0000-0002-3029-6717)
- Arko Lucieer (ORCID: https://orcid.org/0000-0002-9468-4516)
- Megan M. Lewis (ORCID: https://orcid.org/0000-0003-1203-6281)
- Shaun R. Levick (ORCID: https://orcid.org/0000-0003-4437-9174)
Institutions
- Commonwealth Scientific and Industrial Research Organisation (AU)
- University of Tasmania (AU)
- University of Freiburg (DE)
- The University of Adelaide (AU)
- Leipzig University (DE)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-04
- DOI
- https://doi.org/10.5281/zenodo.23946672
- Primary Topic
- Remote Sensing and LiDAR Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00