https://github.com/meetps/tf-3dgan

Tensorflow implementation of 3D Generative Adversarial Network.
https://github.com/meetps/tf-3dgan

deep-learning generative-adversarial-network tensorflow

Last synced: 5 days ago
JSON representation

Tensorflow implementation of 3D Generative Adversarial Network.

• Host: GitHub
• URL: https://github.com/meetps/tf-3dgan
• Owner: meetps
• License: mit
• Created: 2017-02-21T04:31:53.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2019-01-03T20:21:23.000Z (about 6 years ago)
• Last Synced: 2024-12-25T19:05:47.000Z (19 days ago)
• Topics: deep-learning, generative-adversarial-network, tensorflow
• Language: Python
• Homepage: https://meetshah.dev/gan/deep-learning/tensorflow/visdom/2017/04/01/3d-generative-adverserial-networks-for-volume-classification-and-generation.html
• Size: 80.1 KB
• Stars: 284
• Watchers: 17
• Forks: 79
• Open Issues: 18
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# tf-3dgan

[![license](https://img.shields.io/github/license/mashape/apistatus.svg)](https://github.com/meetshah1995/tf-3dgan/blob/master/LICENSE)

[![arXiv Tag](https://img.shields.io/badge/arXiv-1610.07584-brightgreen.svg)](https://arxiv.org/abs/1610.07584)

## Tensorflow implementation of 3D Generative Adversarial Network.

This is a tensorflow implementation of the paper "Learning a Probabilistic Latent Space of Object Shapes 

via 3D Generative-Adversarial Modeling"

![](http://3dgan.csail.mit.edu/images/model.jpg)

[Blog Post with interactive volume plots](https://meetshah1995.github.io/gan/deep-learning/tensorflow/visdom/2017/04/01/3d-generative-adverserial-networks-for-volume-classification-and-generation.html)

### Requirements

* tensorflow>=1.0

* visdom>=1.0.1 (for mesh visualization)

* scipy

* scikit-image

* stl (optional)

#### One-line installation

    

`pip install scipy scikit-image stl visdom`

### Data

* Download the training data from the 3D Shapenet [website](http://3dshapenets.cs.princeton.edu/3DShapeNetsCode.zip)

* Extract the zip and modify the path appropriately in `dataIO.py`

### Usage

Launch [visdom](https://github.com/facebookresearch/visdom#launch) by running

```

python -m visdom.server

```

To train the model (visdom will show generated chairs after every 200 minibatches)

```

python 3dgan_mit_biasfree.py 0 

```

To generate chairs

```

python 3dgan_mit_biasfree.py 1 

```

Some sample generated chairs

|            |              |            |          |          |

|------------|--------------|------------|----------|----------|

|![](http://homepages.iitb.ac.in/~meetshah1995/files/2_.png) | ![](http://homepages.iitb.ac.in/~meetshah1995/files/4_.png) |  ![](http://homepages.iitb.ac.in/~meetshah1995/files/1_.png) |  ![](http://homepages.iitb.ac.in/~meetshah1995/files/3_.png) |  ![](http://homepages.iitb.ac.in/~meetshah1995/files/5_.png) |

### Source code files

| File      | Description                                                                   |

|-----------|-------------------------------------------------------------------------------|

|3dgan_mit_biasfree.py      | 3dgan as mentioned in the paper, with same hyperparams. 

|3dgan.py                   | baseline 3dgan with fully connected layer at end of discriminator.

|3dgan_mit.py               | 3dgan as mentioned in the paper with bias in convolutional layers.

|3dgan_autoencoder.py       | 3dgan with support for autoencoder based pre-training.

|3dgan_feature_matching.py  | 3dgan with additional loss of feature mathcing of last layers. 

|dataIO.py                  | data input output and plotting utilities.

|utils.py                   | tensorflow utils like leaky_relu and batch_norm layer.

### Todo

* Host the trained models

* Add argparser based interface

* Add threaded dataloader

* Release the pytorch and keras versions of the GAN.

* Train for longer number of epochs to improve quality of generated chairs.

### Contributors

* @meetshah1995

* @khushhallchandra