Skip to content

Ricardo-L-C/ColorizationWithRegion

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 

Repository files navigation

ColorizationWithRegion

The code for pg2021 paper "Line Art Colorization Based on Explicit Region Segmentation"

This is a simple implementation for comparison with Tag2Pix (code / paper).

Overall project refactoring and further optimization may be later.

Usage

  1. Build the environment and dataset according to Tag2Pix.

  2. Use python code/skeleton/line_art2skeleton.py <line art folder> to create skeleton maps.

    See DanbooRegion for environment.

    Some code and pretrained model are from DanbooRegion.

    For each line art folders, e.g., keras_train, xdog_train, keras_test or others, create a corresponding folder to place skeleton maps, like keras_train_skeleton and others.

  3. Replace loader/dataloader.py of Tag2Pix with code/loader/dataloader.py to load skeleton maps.

    We also remove the random_jitter for visible test results while training.

  4. For dual-branch, replace network.py and tag2pix.py of Tag2Pix with files in code/dual_branch.

  5. For direct concatenation, replace network.py and tag2pix.py of Tag2Pix with files in code/direct.

  6. Train the model as Tag2Pix.

About

The code for pg2021 paper "Line Art Colorization Based on Explicit Region Segmentation"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages