Comic OCR

An OCR system fine-tuned for comics and manga

Usage

pip install comic-ocr

import comic_ocr

# paragraphs: List[comic_ocr.typing.Paragraph]
paragraphs = comic_ocr.read_paragraphs('../example/xkcd_100.png')

...
paragraphs[0].text # 'Just wrote a paper disproving gravity.'
paragraphs[0].location # Rect(size=(168, 40), at=(15, 32))

Development

The project uses Git LFS to store the default trained models.

# After you normal git clone; Go to the directory and run the following command:
git lfs install
git lfs pull

The project uses uv (pyproject.toml) for dependency management and development:

# To install and sync dependencies
uv sync

# To run tests
uv run pytest

# To run build
uv build