Style Transfer Learning

Style Transfer Learning

In this project, I take the style of Vincent van Gogh's Starry Night and apply it to this city skyline!

This project involves the use of PyTorch to load and retrain the famous VGG-19 Convolutional Neural Network. This is a vastly complex and deep neural network that has proven to be able to find the most detailed of features.


I use this model as a feature extraction tool on the Starry Night image.

Once the model is well-trained, I am able to begin to incrementally begin to apply Starry Night "filters" to the city skyline. This sort of learning is fascinating and the possible uses are endless. I think it will be really interesting to see what Hollywood can do with this sort of technology!

Here is the final image. It took roughly 10,000 iterations of training to get this level of style transfer. I took a snapshot of the model every 300 iterations, combining them into a video. To see the video of the model iteratively training, I have posted a link below in the references section!

References

Code:
https://github.com/zbloss/Style-Transfer/



Model training output video:
https://youtu.be/aWas9mFsPyo/