aipixel / GTNet

[IJCV 2023] The official repo for “Learning Geometric Transformation for Point Cloud Completion”

*GTNet *

[GTNet]

This repository contains the source code for the paper Learning Geometric Transformation for Point Cloud Completion

Installation

Moreover, this repository introduces an integrated Point Cloud Completion Benchmark implemented in Python 3.8, PyTorch 1.8 and CUDA 11.4.

1. Install dependencies:

• h5py 3.9.0
• opencv-python 4.7.0
• matplotlib 3.7.1
• transforms3d 0.4.1
• munch 3.0.0
• open3d 0.13.0
• PyTorch 1.8.0
• PyYAML 5.4.1

2. Compile PyTorch 3rd-party modules (ChamferDistancePytorch, emd, expansion_penalty, MDS, Pointnet2.PyTorch)

MVP Dataset

Please download the MVP Dataset (Dropbox or Google Drive) to the folder data.

Usage

• To train a model: run python train.py -c *.yaml, e.g. python train.py -c pcn.yaml
• To test a model: run python test.py -c *.yaml, e.g. python test.py -c pcn.yaml
• Config for each algorithm can be found in cfgs/.
• run_train.sh and run_test.sh are provided for SLURM users.

[Citation]

If you find our code useful, please cite our paper:

@article{zhang2023learning,
  title={Learning Geometric Transformation for Point Cloud Completion},
  author={Zhang, Shengping and Liu, Xianzhu and Xie, Haozhe and Nie, Liqiang and Zhou, Huiyu and Tao, Dacheng and Li, Xuelong},
  journal={International Journal of Computer Vision},
  pages={1--21},
  year={2023},
  publisher={Springer}
}

[License]

Our code is released under MIT License.

[Acknowledgement]

Our code is inspired by GRNet and VRCNet. We thank the authors for their great job!