Implementing of a Neural Algorithm of Artistic Style

This is an implementation of the "A Neural Algorithm of Artistic Style". This uses the selected filtered responses of a pre-trained model (VGG-19) to capture low level to high level features and transfer them to the content image.

You can now do it through the IPython Notebook available here!

This code is documented so you can follow along while reading with the paper. You can also just replace with your own images and to generate your new painting.

Results:

Using the StarryNight:

How to run

You will need to install dependencies:

• TensorFlow
• Scipy
• Numpy

You will need to download the VGG-19 model.

Then just run art.py.

References:

• A Neural Algorithm of Artistic Style
• https://github.com/jcjohnson/neural-style
• https://github.com/ckmarkoh/neuralart_tensorflow