https://github.com/diegovalsesia/GraphCNN-GAN-codeonly

Graph-convolutional GAN for point cloud generation. Code from ICLR 2019 paper Learning Localized Generative Models for 3D Point Clouds via Graph Convolution. Code only.
https://github.com/diegovalsesia/GraphCNN-GAN-codeonly

Last synced: 2 months ago
JSON representation

Graph-convolutional GAN for point cloud generation. Code from ICLR 2019 paper Learning Localized Generative Models for 3D Point Clouds via Graph Convolution. Code only.

• Host: GitHub
• URL: https://github.com/diegovalsesia/GraphCNN-GAN-codeonly
• Owner: diegovalsesia
• Created: 2019-07-23T08:39:47.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2019-07-23T08:41:42.000Z (over 5 years ago)
• Last Synced: 2024-08-01T03:45:59.883Z (5 months ago)
• Language: Python
• Size: 24.4 KB
• Stars: 13
• Watchers: 0
• Forks: 3
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Learning Localized Generative Models for 3D Point Clouds via Graph Convolution (ICLR 2019)

Code-only repository. Full repository with trained models (large files!): https://github.com/diegovalsesia/GraphCNN-GAN

If you like our work, please cite the journal version of the paper.

Journal version BibTex reference:

```

@ARTICLE{Valsesia2019journal,

author={Diego {Valsesia} and Giulia {Fracastoro} and Enrico {Magli}},

journal={under review},

title={Learning Localized Representations of Point Clouds with Graph-Convolutional Generative Adversarial Networks},

year={2019},

volume={},

number={},

pages={},

}

```

ICLR 2019 BibTex reference:

```

@inproceedings{valsesia2019learning,

  title={Learning Localized Generative Models for 3D Point Clouds via Graph Convolution},

  author={Valsesia, Diego and Fracastoro, Giulia and Magli, Enrico},

  booktitle={International Conference on Learning Representations (ICLR) 2019},

  year={2019}

}

```

# Requirements

  - Python 2.7

  - Tensorflow >=1.6 

# Usage

  

  A trained model for the method with aggregation upsampling is provided for the following Shapenet classes: airplane, chair, sofa, table.

  - launch_test.sh : generate a batch of point clouds from the specified class 

  - launch_train.sh : retrain the network (requires downloading the Shapenet dataset and place it in the data directory)