AGIS-Net

[SIGGRAPH Asia 2019] Artistic Glyph Image Synthesis via One-Stage Few-Shot Learning

View project on GitHub

Data preparation

Convert font files to images

Please use the following code: font2image

Prepocess the glyph images

Every glyph image is arranged as follow:

A: Base image, B: Shape image, C: Texture image

code example: generate_dataset_example.py

Prepare the pre-train dataset

xyz_dataset/
    train/
        XXX_XXX.png
        XXX_XXX.png
        ...
    val/
        XXX_XXX.png
        XXX_XXX.png
        ...
    test/
        XXX_XXX.png
        XXX_XXX.png
        ...

Prepare the fine-tune datasets

abc_dataset/
    few_dict.txt        few-shot reference glyphs
    style/
        XXX_1.png
        XXX_2.png
        ...
    train/
        XXX_XXX.png
        XXX_XXX.png
        ...
    val/
        XXX_XXX.png
        XXX_XXX.png
        ...

Please read the xxx_dataset.py files, you can get more details.

Chinese average font style(skeleton)

The average skeleton glyph image dataset including 6763 common used characters’ glyph images with size 320X320. You can also download it via the download_dataset.sh script.