style-transfer

style-transfer copied to clipboard

style-transfer

style transfer code for generating stylized images using content and style loss from different layers, from different loss networks (vgg-16, vgg-19, inception-v1, inception-v2, inception-v3 trained on imagenet, inception-v3 trained on openimages, inception-v4)

Requirements

tensorflow r1.3

Documentation

Training Style Transfer Networks

./setup.sh
python fast_style.py --test-path data/content/ --train-path data/train/ --result-path test-fast/ --style-path data/style/wave.jpg

To access all command line arguments, run python fast_style.py --help

Stylizing a single image

python slow_style.py --style-path data/style/wave.jpg --content-path data/content/stata.jpg --result-path test-slow/

To access all command line arguments, run python slow_style.py --help

Printing model layers with tensor sizes

python print_model_layers.py --network inception-v4

To access all command line arguments, run python print_model_layers.py --help

Roadmap

• [ ] Add explanatory comments
• [ ] Expose more command-line arguments

Contributing

Please feel free to:

• Create an issue
• Open a Pull Request
• Join the gitter chat
• Share your success stories!