/koori

Mokuro fork intended to aid manga scanlation efforts.

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

Koori

A fork of Yuki which is in turn a fork of Mokuro. Automatically OCRs manga pages and outputs the results to a TXT file.

Requires Python 3.10 or higher.

Usage

  1. Clone the repository by using git or downloading the zip
  2. Install the required packages with pip install -r requirements.txt
  3. Place pages in the images folder
  4. Run main.py with python main.py

The outputs will have the same name as the pages and will be exported in the text folder. There is also a file named combined_text.txt that combines all the outputs into one with page headers.

Note

Remember that the OCR is not 100% accurate and you will still need to manually correct mistakes. This simply exists to speed up the process.

To-do

  • Combine all the outputs into a single file, with headers signifying which sections are from
  • Have the OCR follow the text in "reading order" (This is probably complicated as hell to do, so it's just here as an "idea" of sorts.)