Repolode
Submit
zyddnys

zyddnys/manga-image-translator

Open-source tool that detects, OCRs, translates, erases and re-typesets text in manga and other images, mainly Japanese to Chinese or English.

View on GitHubWebsite
Stars
10,519
Forks
1,078
Open issues
144
Last commit
10 days ago

What is it?

Manga Image Translator is a Python pipeline for translating text inside images such as manga pages, comics and screenshots. It chains text detection, OCR, machine translation, inpainting to remove the original text, and rendering of the translated text back onto the image. It mainly targets Japanese, and also supports Simplified and Traditional Chinese, English and a number of other languages.

The project can be run as a local batch tool over a folder of images, as a web server with a browser UI, as an API service, or through Docker. Translation can use online services such as DeepL, Baidu, Youdao or OpenAI-compatible APIs, or offline models such as NLLB, M2M100 and Sugoi. Users can tune detection, inpainting, font and rendering options through command-line flags or a JSON configuration file.

The README notes the project is in early development. The hosted demo is no longer working, so self-hosting is required.

Key features

  • 01Detects text regions in manga and images, then runs OCR on them.
  • 02Removes original text through inpainting and re-renders the translation.
  • 03Supports many translators, including online APIs and offline models such as NLLB and Sugoi.
  • 04Offers local batch mode, a web UI, an API server and Docker images.
  • 05Lets users tune detection, upscaling, fonts and colorization via flags or a config file.

Use cases

  • Reading Japanese manga or doujin pages that have no official translation.
  • Batch-translating a folder of comic or screenshot images for personal reading.
  • Preparing translated images for manual typesetting by exporting cleaned, inpainted backgrounds.
  • Running an in-house image translation service through its API or web server.

Quick start

  1. 1

    Ensure Python 3.12+ is installed, then clone the repository

    git clone https://github.com/zyddnys/manga-image-translator.git
  2. 2

    Create a virtual environment

    python -m venv venv
  3. 3

    Activate the virtual environment (Linux/macOS)

    source venv/bin/activate
  4. 4

    Install dependencies from requirements.txt

    pip install -r requirements.txt

Things to know

  • The README says the project is in early development with many shortcomings.
  • Needs a C++ compiler and Git to build a dependency, and Windows users need Microsoft C++ Build Tools.
  • Docker image is about 15GB, and GPU use requires PyTorch matched to your CUDA version.
  • Licensed under GPL-3.0, a copyleft license.

Topics

#manga#ocr#image-translation#machine-translation#inpainting#text-detection#pytorch#computer-vision#anime#auto-translation

FAQ

What is manga-image-translator?
Open-source tool that detects, OCRs, translates, erases and re-typesets text in manga and other images, mainly Japanese to Chinese or English. Manga Image Translator is a Python pipeline for translating text inside images such as manga pages, comics and screenshots. It chains text detection, OCR, machine translation, inpainting to remove the original text, and rendering of the translated text back onto the image. It mainly targets Japanese, and also supports Simplified and Traditional Chinese, English and a number of other languages.
Is manga-image-translator free to use? What license does it have?
manga-image-translator is open source under the GPL-3.0 license. Check the license terms before using it in commercial products.
Is manga-image-translator still maintained?
Actively maintained. The last commit to manga-image-translator was on Oct 11, 2026.
What can I use manga-image-translator for?
Reading Japanese manga or doujin pages that have no official translation. Batch-translating a folder of comic or screenshot images for personal reading. Preparing translated images for manual typesetting by exporting cleaned, inpainted backgrounds. Running an in-house image translation service through its API or web server.
How do I install manga-image-translator?
Ensure Python 3.12+ is installed, then clone the repository: git clone https://github.com/zyddnys/manga-image-translator.git Create a virtual environment: python -m venv venv Activate the virtual environment (Linux/macOS): source venv/bin/activate Install dependencies from requirements.txt: pip install -r requirements.txt
Where can I get manga-image-translator?
The source code and installation instructions are on GitHub: https://github.com/zyddnys/manga-image-translator

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects