Mesh Neural Networks Based on Dual Graph Pyramids

The repository contains official jittor implementations for DGNet.

The paper is in Here.

Installation

pip install -r requirements.txt

Scannet

Download the scannet and prepare it.

Change tools/process_scannet.py "data_dir"

bash run_prepare.sh

Models

We release our trained models and training logs in "work_dirs".

Evaluation

To evaluate the model, run:

bash run.sh 0 configs/tvcg_scene_scannetv2_val.py --task=val_iters
# for test 
bash run.sh 0 configs/tvcg_scene_scannetv2_test.py --task=val_iters

For final prediction:

# change the dir before merging.
python3 tools/merge_results.py

Train

bash dist_run.sh 0,1,2,3 4 configs/tvcg_scene_scannetv2_train.py --task=train

Citation

If you find our repo useful for your research, please consider citing our paper:

@article{li2023mesh,
  title={Mesh neural networks based on dual graph pyramids},
  author={Li, Xiang-Li and Liu, Zheng-Ning and Chen, Tuo and Mu, Tai-Jiang and Martin, Ralph R and Hu, Shi-Min},
  journal={IEEE Transactions on Visualization and Computer Graphics},
  year={2023},
  publisher={IEEE}
}

LICENSE

This repo is under the Apache-2.0 license. For commercial use, please contact the authors.

li-xl/DGNet