manga-image-translator

Why we included this project

Manga and webcomic readers who are stuck with untranslated image-heavy content will find this tool does the full job: it finds the text, reads it with OCR, translates it, wipes the original, and sets the translation back into the art. Japanese is the main language, with Simplified and Traditional Chinese, English, and roughly twenty more languages covered, so it matches what actually shows up on image boards and scanlations. You can run it as a batch CLI, a local web interface, or an API; Docker and a Colab notebook lower the setup barrier for people who avoid Python. The OCR, text detection, and inpainting stages are also cleanly separated, which makes the codebase a useful reference if you're building your own image-translation pipeline. Just know it's early-stage and the public demo is gone, so plan to run it yourself and adjust translation settings for your material.