/Artistic-Transfer

Text based style transfer on segmented images

Primary LanguageJupyter Notebook

Artistic Transfer

This is the implementation of the paper Segment-Based, User-Generated Image Styling with Neural Style Transfer

What does it do?

  • Generate style images from a prompt provided by the user
  • Train a Transformation network on the generated style image
  • Apply style transfer on one specific segment from the content image

Here's a glimpse of how it works: flowchart

Here are some sample results image

Note: Before playing around with the code, make sure to add your own hugging face token in the first cell.