Mondrian Cat

Inspiration

I read a tutorial on neural style transfer on the official TensorFlow website and thought about using this technique to turn an image into a colorful abstract painting.

What it does

This project takes in a image of a cat and a 1921 painting by Piet Mondrian and uses neural style transfer with TensorFlow to create an image that looks like the cat image in the style of the painting.

How we built it

I used TensorFlow, Matplotlib, Numpy, a few other Python libraries, and a tutorial (https://www.tensorflow.org/tutorials/generative/style_transfer) to achieve style transfer.

Challenges we ran into

I found that the times for running each epoch was surprisingly long; I used fewer epochs with fewer steps for testing purposes.

Accomplishments that we're proud of

This is my first time using style transfer techniques.

What we learned

I learned how to achieve neural style transfer to generate a wild-looking image.

What's next for Mondrian Cat

I plan to increase the number of epochs and the number of steps per epoch to create an even more interesting picture.

Built With

• google-colaboratory
• matplotlib
• numpy
• python
• tensorflow