/Artify

:art: Keras implementation of Neural Style Transfer for coloring paintings and creating art with playground

Primary LanguagePython

Artify is Neural Style Transfer playground for creating art, coloring the sketches, and transfering the styles to content image. It is implemented to faster experimentation with different images and settings.

Requirements

  • PyQt5
  • Keras
  • numpy
  • scipy
  • h5py
  • scikit-image

Run

Following code-snippet opens window to select content and style images. It gives you flexibility to change the hyperparameter of the learning process.

python3 gui.py

After generation of style transfered images, if you want to preserve the original color of the content image then:

python3 color_transform.py <path-to-content-image> <path-to-generated image>

NOTE: Copy results, after generating art, from result directory.

Results

Content Style Art color preserved

NOTE: All results will be added soon! Above are initial results.

Todo

  • Colab support
  • custom model support for finding better settings
  • preserve color in GUI

License

MIT

Author

Ghanshyam_Chodavadiya

Acknowledgement

💚 ozamanan