FuelDeep3D: 3D Fuel Segmentation Using Terrestrial Laser Scanning and Deep
Learning
Provides tools for preprocessing, feature extraction, and segmentation
of three-dimensional forest point clouds derived from terrestrial laser scanning.
Functions support creating height-above-ground (HAG) metrics, tiling, and sampling
point clouds, generating training datasets, applying trained models to new point
clouds, and producing per-point fuel classes such as stems, branches, foliage,
and surface fuels. These tools support workflows for forest structure analysis,
wildfire behavior modeling, and fuel complexity assessment. Deep learning
segmentation relies on the PointNeXt architecture described by Qian et al.
(2022) <doi:10.48550/arXiv.2206.04670>, while ground classification utilizes
the Cloth Simulation Filter algorithm by Zhang et al. (2016) <doi:10.3390/rs8060501>.
| Version: |
0.1.1 |
| Depends: |
R (≥ 4.1) |
| Imports: |
stats, RColorBrewer, viridisLite, rlang |
| Suggests: |
lidR, reticulate, dbscan, ggplot2, rgl, RCSF, scales |
| Published: |
2026-03-02 |
| DOI: |
10.32614/CRAN.package.FuelDeep3D (may not be active yet) |
| Author: |
Venkata Siva Reddy Naga [aut, cre],
Alexander John Gaskins [aut],
Carlos Alberto Silva [aut] |
| Maintainer: |
Venkata Siva Reddy Naga <venkatasivareddy003 at gmail.com> |
| BugReports: |
https://github.com/venkatasivanaga/FuelDeep3D/issues |
| License: |
GPL (≥ 3) |
| URL: |
https://github.com/venkatasivanaga/FuelDeep3D |
| NeedsCompilation: |
no |
| Materials: |
README |
| CRAN checks: |
FuelDeep3D results |
Documentation:
Downloads:
Linking:
Please use the canonical form
https://CRAN.R-project.org/package=FuelDeep3D
to link to this page.