TreeNet3D is the first large-scale, multi-type synthetic tree model dataset. The dataset comprises 13,000 models of ten common tree species: camphor tree, small-leaved olive, agarwood, flooded gum, lemon tree, lombardy poplar ,tibetan cherry, flamboyant tree, fir tree and ajang olive, with each species represented by 1,300 samples. Each tree model offers a 3D point cloud with hierarchical structure and precise parameters, along with information on tree skeletons and volume. The 3D point cloud is divided into original and noise-added variants, with a distinct separation structure for branches and leaves. Each branch is differentiated and assigned a branch number attribute based on its sequence. The tree skeleton abstracts the tree branches into node information, with each skeleton node containing hierarchical relationships and accurate tree radius information.
If you would like to download the TreeNet3D data, please fill out an agreement to the TreeNet3D Terms of Use, using your institutional email addresses, and send it to us at shengjuntang@szu.edu.cn.
The directory has the following structure:
|-- Flamboyant Tree
| |-- Flamboyant Tree00_01
| | |-- meshmodel
| | | |-- leavesmesh.obj
| | | |-- trunkmesh.obj
| | |-- pointcloud
| | | |-- branchPointClouds.txt
| | | |-- leafPointClouds.txt
| | | |-- mergePointClouds.txt
| | |-- pointcloudnoise
| | | |-- branchPointClouds.txt
| | | |-- leafPointClouds.txt
| | | |-- mergePointClouds.txt
| | |-- skeleton
| | | |-- skeleton.txt
| | |-- TreeVolum.txt
| |-- Flamboyant Tree00_02
| |-- ...
|-- Small-leaved Olive
|-- Camphor Tree
|-- ...
The following are overviews of the data formats used in TreeNet3D:
*.obj: Mesh in various parts of trees.
*PointClouds.txt: Point clouds in various parts of trees. Contains attributes as follows:
[ X, Y, Z, Point Cloud-Branch Index, Point Cloud-Branch Hereditary Relations]
*skeleton.txt: tree skeleton points. Contains attributes as follows:
[ X, Y, Z, Skeleton-Radius, Skeleton-Branch Hereditary Relations, Skeleton-Branch Index]
If you use the TreeNet3D data please cite:
@article{tang2024treenet3d,
title={TreeNet3D: A large scale tree benchmark for 3D tree modeling, carbon storage estimation and tree segmentation},
author={Tang, Shengjun and Ao, Zhuoyu and Li, Yaoyu and Huang, Hongsheng and Xie, Linfu and Wang, Ruisheng and Wang, Weixi and Guo, Renzhong},
journal={International Journal of Applied Earth Observation and Geoinformation},
volume={130},
pages={103903},
year={2024},
publisher={Elsevier}
}
The data is released under the TreeNet3D Terms of Use, and the code is released under the MIT license.
If you have any questions, please contact us at shengjuntang@szu.edu.cn