Purpose

What YomiNinja does

YomiNinja is an OCR overlay application for Japanese language learners. It captures on-screen text from any application — games, visual novels, video players, browsers — using optical character recognition, overlays that text directly on top of the source window, and connects it to Yomitan's popup dictionary.

The result: hover any Japanese word in any game and get an instant definition, reading, and pitch accent popup — without alt-tabbing, without pausing, without breaking your immersion.

It works on anything where Japanese text is visible on screen. No game engine compatibility requirements. No process injection. No reverse engineering. Just screen capture, OCR, and a transparent overlay.

Author

The developer

matt-m-o

github.com/matt-m-o

YomiNinja was created and is maintained by matt-m-o, a developer and Japanese learner who built the tool to solve their own immersion workflow problem: playing Japanese games while being able to look up unfamiliar vocabulary without constantly interrupting the game.

The project launched on September 8, 2023 and reached its first stable v0.9.x series in August 2024, growing from a personal tool into a community-supported application used by hundreds of language learners worldwide.

Development is sustained by Patreon supporters — 934 members as of 2024, with 66 paying supporters providing monthly funding. Patreon supporters get early access to pre-release builds.

"The goal was simple: be able to hover over any kanji in any game and get a dictionary popup, without any alt-tabbing. Everything else followed from that."

History

Key milestones

Sep 2023

Initial public release

YomiNinja launches on GitHub as an open-source project. Core feature: OCR capture with Yomitan popup dictionary integration.
Early 2024

Linux & cloud OCR support

Linux AppImage released. Google Cloud Vision and Google Lens added as cloud OCR backends. Character-level text positioning enables precise hover detection.
Jul 2024

macOS & MangaOCR

macOS support arrives with Apple Vision Framework integration. MangaOCR added for stylized fonts. Auto OCR and OCR Templates introduced — enabling hands-free, continuous text detection.
Aug 2024

v0.9.1 — major feature release

VOICEVOX TTS integrated across all platforms. GPU acceleration via CUDA and ROCm. MangaOCR on-demand mode. Furigana threshold control. Faster screenshots. Browser overlay / PWA mode.
Aug 4, 2024

v0.9.3 — current stable release

Bug fixes for Google Cloud Vision, custom shortcuts, and macOS display scaling. The current recommended version for all platforms.

Under the hood

The technology

YomiNinja is built on a modern, cross-platform stack designed for low latency and extensibility.

TypeScript + Electron

The application shell is written in TypeScript and runs on Electron, providing cross-platform desktop capabilities and a Chromium browser context for dictionary extension integration.

gRPC backend services

OCR engines and backend processing communicate via gRPC, a high-performance RPC framework. This architecture allows different OCR engines to be added or swapped without core changes.

PaddleOCR & MangaOCR

Local OCR engines powered by deep learning models. PaddleOCR handles general Japanese text; MangaOCR is fine-tuned on manga and comic fonts for superior accuracy on stylized characters.

Transparent overlay

A transparent Electron window sits above the game in the display stack. Extracted text is positioned pixel-accurately to match on-screen text locations, enabling precise hover detection.

Yomitan integration

Yomitan and 10ten Reader are embedded Chrome extensions within YomiNinja's Electron Chromium context. When you hover overlay text, the extension triggers exactly as it would in a browser.

WebSocket output

Every recognized text result is broadcast over a local WebSocket server, enabling integration with external tools: texthooker pages, Anki mining workflows, or custom applications.

License

Open source & free forever

YomiNinja is released under the GNU General Public License v3.0 (GPL-3.0). This means:

The source code is public and auditable by anyone
You can modify and redistribute it under the same license
No proprietary components — the full stack is inspectable
Free to use with no feature gates, subscriptions, or expiry

View on GitHub Report a bug

Keep it going

Support the project

YomiNinja is free to use and will always remain so. If it has saved you hours of frustration looking up kanji, consider supporting the developer on Patreon.

Patreon

Monthly support. Includes early access to pre-release builds before they are publicly released.

Pre-release builds
Supporter-only updates
Direct influence on feature priority

Support on Patreon

Ko-fi

One-time or recurring contributions. No tiers, no commitment — just a direct way to say thanks.

One-time or monthly
No account required for one-time
All contributions go to development

Buy a coffee

Other ways to contribute

Report bugs — open an issue on GitHub Issues with a clear description and steps to reproduce
Share with the community — mention YomiNinja in Japanese learning subreddits, Discord servers, and forums where it might help others
Contribute code — pull requests are welcome for bug fixes, OCR engine integrations, and accessibility improvements

About YomiNinja