Neural Style transfer is a cool application of convolutional neural networks that combines two images into one. Specifically, the style of one image is transfered to the content of the second image.
Neural Style Transfer is based on Gatys et al. (2015). Two images are iteratively merged by transfering the style of one to the content of the other. The process is based on convolutional neural networks and uses a content cost and a style cost to transfer the visual style.
This implementationn of neural style transfer in Tensorflow/Keras is based on the deeplearning.ai Coursera course on Convolutional Neural Networks and the Keras code example.
I used a pre-trained VGG19 neural network (Simonyan & Zisserman, 2014).
Example images to try style transfer and example generated images cam be found in the respective folders. I particularly like the mosaique which gives striking results after only around 40 iterations.
The code takes two images (a content and a style image) and generates a new image. The script can be executed from the command line using:
python3 style_transfer.py <content_image_path> <style_image_path>
The generated image will be optimized for 140 iterations, generating an image every 5 steps.
The output folder can be specified using the --out_dir flag.
An example can be run using python3 style_transfer.py --demo.
For a complete description of commands, use
python3 style_transfer.py -h
To adjust the hyperparameters or number of iterations, feel free to edit style_transfer.py itself.
[1] deeplearning.ai Coursera specialization: Convolutional Neural Networks -
ArtTrans
[2] Keras Neural Style transfer example