https://github.com/matthias-wright/siamese_architecture

PyTorch implementation of the siamese architecture for style transfer I developed for my master's thesis.
https://github.com/matthias-wright/siamese_architecture

autoencoder deep-learning generative-adversarial-network python pytorch style-transfer

Last synced: 3 days ago
JSON representation

PyTorch implementation of the siamese architecture for style transfer I developed for my master's thesis.

• Host: GitHub
• URL: https://github.com/matthias-wright/siamese_architecture
• Owner: matthias-wright
• Created: 2019-11-23T12:33:30.000Z (about 5 years ago)
• Default Branch: master
• Last Pushed: 2021-05-12T11:56:21.000Z (almost 4 years ago)
• Last Synced: 2024-12-14T20:07:50.811Z (about 2 months ago)
• Topics: autoencoder, deep-learning, generative-adversarial-network, python, pytorch, style-transfer
• Language: Python
• Homepage:
• Size: 27.1 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# A Siamese Architecture for Neural Style Transfer

The implementation of the siamese architecture that I developed for my master's thesis.

## Dependencies

* Python 3.5.2+

* NumPy 1.11.0+

* PyTorch 0.4.1+

* PIL 1.1.7+

* Matplotlib 3.0.1+

## Example



  



## Instructions

### Training

```python

import siamese

data_photo = '...' # file path to the photograph images (RGB) folder

data_oil = '...' # file path to the oil images (RGB) folder

siam_model = siamese.Siamese(in_path_photo=data_photo,

                             in_path_oil=data_oil, 

                             autoencoder_path='saves/autoencoders', 

                             num_epochs=100, batch_size=50,

                             learning_rate=0.0002, 

                             recon_loss_weight=10, 

                             penalty_coef=10, verbose=True)

                          

siam_model.train()

```

### Load Pretrained Model

```python

import siamese

data_photo = '...' # file path to the photograph images (RGB) folder

data_oil = '...' # file path to the oil images (RGB) folder

siam_model = siamese.Siamese(in_path_photo=data_photo,

                             in_path_oil=data_oil, 

                             autoencoder_path='saves/autoencoders', 

                             num_epochs=100, batch_size=50,

                             learning_rate=0.0002, 

                             recon_loss_weight=10, 

                             penalty_coef=10, verbose=True)

siam_model.load(epoch=18, path='saves')                                 

```

### Style Transfer

```python

import siamese

import util

siam_model = siamese.Siamese(autoencoder_path='saves/autoencoders')

siam_model.load(epoch=18, path='saves')     

x_photo = util.image_to_tensor('image.jpg')

x_oil = siam_model.translate_photo_to_oil(x_photo)

x_oil = util.tensor_to_numpy(x_oil)

x_oil = util.numpy_to_image(x_oil)

x_oil.save('stylized_image.png')                                    

```

## License

This project is licensed under the [MIT Licence](https://choosealicense.com/licenses/mit/)