https://github.com/yunjey/domain-transfer-network

TensorFlow Implementation of Unsupervised Cross-Domain Image Generation
https://github.com/yunjey/domain-transfer-network

domain-transfer image-generation tensorflow unsupervised-learning

Last synced: 15 days ago
JSON representation

TensorFlow Implementation of Unsupervised Cross-Domain Image Generation

• Host: GitHub
• URL: https://github.com/yunjey/domain-transfer-network
• Owner: yunjey
• License: mit
• Created: 2016-12-30T08:00:26.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2018-06-06T08:24:49.000Z (over 6 years ago)
• Last Synced: 2024-08-28T18:14:00.021Z (3 months ago)
• Topics: domain-transfer, image-generation, tensorflow, unsupervised-learning
• Language: Python
• Homepage:
• Size: 7.39 MB
• Stars: 861
• Watchers: 53
• Forks: 204
• Open Issues: 10
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Domain Transfer Network (DTN) 

TensorFlow implementation of [Unsupervised Cross-Domain Image Generation.](https://arxiv.org/abs/1611.02200)

![alt text](jpg/dtn.jpg)

## Requirements

* [Python 2.7](https://www.continuum.io/downloads)

* [TensorFlow 0.12](https://github.com/tensorflow/tensorflow/tree/r0.12)

* [Pickle](https://docs.python.org/2/library/pickle.html)

* [SciPy](http://www.scipy.org/install.html)




## Usage

#### Clone the repository

```bash

$ git clone https://github.com/yunjey/dtn-tensorflow.git

$ cd dtn-tensorflow

```

#### Download the dataset

```bash

$ chmod +x download.sh

$ ./download.sh

```

#### Resize MNIST dataset to 32x32 

```bash

$ python prepro.py

```

#### Pretrain the model f

```bash

$ python main.py --mode='pretrain'

```

#### Train the model G and D

```bash

$ python main.py --mode='train'

```

#### Transfer SVHN to MNIST

```bash

$ python main.py --mode='eval'

```




## Results

#### From SVHN to MNIST 

![alt text](jpg/svhn_mnist_2900.gif)

![alt text](jpg/svhn_mnist_2900.png)

![alt text](jpg/svhn_mnist_3700.png)

![alt text](jpg/svhn_mnist_5300.png)

#### From Photos to Emoji (in paper)

![alt text](jpg/emoji_1.png)

![alt text](jpg/emoji_2.png)