TreeFormer: Single-view Plant Skeleton Estimation via Tree-constrained Graph Generation (WACV2025)

Xinpeng Liu¹, Hiroaki Santo¹, Yosuke Toda^2,3, Fumio Okura¹
(¹ Osaka University, ² Phytometrics, ³ Nagoya University)

Requirements

• CUDA>=9.2
• PyTorch>=1.7.1

For other system requirements please follow

pip install -r requirements.txt

Compiling CUDA operators

cd ./models/ops
python setup.py bulid install

Code Usage

1. Dataset preparation

Please download [Guyot dataset] by following the steps under Usage. The structure of the dataset should be as follows:

guyot_data/
└── train/
    └── check/
        └── images
    └── data/
        └── Set02_IMG_3468.pt
    └── img/
        └── images
    └── unet/
        └── images
└── test/
    └── check/
        └── images
    └── data/
        └── Set02_IMG_3468.pt
    └── img/
        └── images
    └── unet/
        └── images
└── val/
    └── check/
        └── images
    └── data/
        └── Set02_IMG_3468.pt
    └── img/
        └── images
    └── unet/
        └── images

2. Training

2.1 Prepare config file

The config file can be found at .configs/tree_2D_use_mst_only1.yaml and .configs/tree_2D_use_unmst_only1.yaml. Make custom changes if necessary.

2.2 Train

For example, the command for training Relationformer is following:

python -m torch.distributed.launch --nproc_per_node=8 train.py --config configs/tree_2D_use_mst_only1.yaml --cuda_visible_device 0 1 2 3 4 5 6 7 

python -m torch.distributed.launch --nproc_per_node=8 train.py --config configs/tree_2D_use_mst_only1.yaml --cuda_visible_device 0 1 2 3 4 5 6 7 --resume trained_weights/check/checkpoint_81_epoch.pkl 

3. Evaluation

Once you have the config file and trained model, run following command to evaluate it on test set:

python valid_smd_guyot_nx.py

4. Citation

@inproceedings{liu2025treeformer,
  title={{TreeFormer}: Single-view Plant Skeleton Estimation via Tree-constrained Graph Generation},
  author={Liu, Xinpeng and Santo, Hiroaki and Toda, Yosuke and Okura, Fumio},
  booktitle={Proceedings of IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
  year={2025}
}