Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/muhammedbuyukkinaci/neural-style-transfer-with-tensorflow

Implementation of Neural Style Transfer algorithm with pre-trained VGG-16 Network & TensorFlow in Python 3.
https://github.com/muhammedbuyukkinaci/neural-style-transfer-with-tensorflow

artistic-style-transfer neural-style-transfer neural-style-transfer-tensorflow tensorflow vgg-16

Last synced: 3 months ago
JSON representation

Implementation of Neural Style Transfer algorithm with pre-trained VGG-16 Network & TensorFlow in Python 3.

Host: GitHub
URL: https://github.com/muhammedbuyukkinaci/neural-style-transfer-with-tensorflow
Owner: MuhammedBuyukkinaci
Created: 2018-06-08T22:03:56.000Z (over 6 years ago)
Default Branch: master
Last Pushed: 2022-11-21T21:28:18.000Z (about 2 years ago)
Last Synced: 2024-10-08T18:21:38.520Z (3 months ago)
Topics: artistic-style-transfer, neural-style-transfer, neural-style-transfer-tensorflow, tensorflow, vgg-16
Language: Python
Homepage:
Size: 14.5 MB
Stars: 4
Watchers: 1
Forks: 3
Open Issues: 3
Metadata Files:
- Readme: README.md
- Changelog: changed_content.jpg

Awesome Lists containing this project

README

# Neural Style Transfer with TensorFlow
Implementation of Neural Style Transfer algorithm with pretrained VGG-16 Network & TensorFlow in Python 3.

# Dependencies

```pip install -r requirements.txt```

```pip3 install -r requirements.txt```

# Pre-trained VGG-16 Network ( Warning : 491 MB)

Download VGG-16 named as ```imagenet-vgg-verydeep-16``` from [here](http://www.vlfeat.org/matconvnet/pretrained/#imagenet-ilsvrc-classification). Put it in _Neural-Style-Transfer-with-TensorFlow_ folder.

![alt text](https://github.com/MuhammedBuyukkinaci/Neural-Style-Transfer-with-TensorFlow/blob/master/pictures/vgg16_where.png)

# Cloning and Training

```
git clone https://github.com/MuhammedBuyukkinaci/Neural-Style-Transfer-with-TensorFlow.git

cd ./Neural-Style-Transfer-with-TensorFlow

python neural_style_transfer.py
```

To train with your parameters, run a command alike below command: (Details in Arguments for Terminal section)

```python neural_style_transfer.py -c_i changed_content.jpg -s_i changed_style.jpg -c_w 25 -s_w 25 -n_i 1000 ```

# Arguments for Terminal
- c_i: Name of Content image. It must be string. Example: changed_content.jpg, default_content.jpg , content_image.png .

- s_i: Name of Style image. It must be string. Example: changed_style.jpg , default_style.jpg , style_image.png .

- c_w: Weight of Content image in loss function. It must be integer. Example: 5 , 20 , 50 .

- s_w: Weight of Style image in loss function. It must be integer. Example: 10 , 20 , 40 .

- s_w: How many iterations to train. It must be integer. Example: 500 , 1000 , 2000 .
# Notebook

Download .ipynb file from [here](https://github.com/MuhammedBuyukkinaci/My-Jupyter-Files/blob/master/neural_style_transfer.ipynb) and run:

```jupyter lab ``` or ```jupyter notebook ```

# Outputs

Content Image | Style Image | Generated Image After 2000 iterations |GIF during 2000 iterations
:-------------------------:|:-------------------------:|:-------------------------:|:-------------------------:
| | |

The last content image belongs to one of my friends, [Umut Başdemir](https://www.linkedin.com/in/umutbasdemir/). He is a Business Development Specialist at Samsung.

# References

- https://www.coursera.org/specializations/deep-learning

- https://www.youtube.com/watch?v=Re2C9INXCNc&index=38&list=PLBAGcD3siRDjBU8sKRk0zX9pMz9qeVxud

- https://gist.github.com/prajjwal1/8bc39c430c8a0303c1430b02207b09f4

- https://github.com/harsha1601/StyleTransfer/blob/master/cnn-st.ipynb

- https://github.com/nestor94/apptist/blob/master/model/style_transfer_model.py.ipynb