Free forever
Download YomiNinja — Free OCR Overlay for Japanese Learners
No account required. No subscription. No data collection. Latest public release: v0.9.3 from August 4, 2024, with platform-specific setup notes below.
Download by platform
Windows
Windows 10 & 11 · Full feature set
System Requirements
| OS | Windows 10 (21H2+) or Windows 11 |
| Required | Visual C++ Redistributables (VCRedist) |
| N/KN editions | Windows Media Feature Pack required |
| GPU (optional) | NVIDIA with CUDA 11.8 for GPU-accelerated OCR |
| RAM | 4 GB minimum; more helps when using heavier OCR engines |
Installation
-
1
Download the installer
Click the button above to go to the latest GitHub release. Download the
.exeinstaller file. -
2
Run the installer
Windows may show a SmartScreen prompt — click More info → Run anyway. YomiNinja is not signed with an expensive certificate, but is fully open source and verifiable on GitHub.
-
3
Launch and configure
Open YomiNinja, import JMdict into Yomitan, select a capture window, choose your OCR engine, and test your first lookup. See the setup guide for the full path.
Linux
AppImage · .deb · .rpm · .pacman · X11
System Requirements
| Display server | X11 required (Wayland: experimental in v0.9.1) |
| Required package | xdotool — install via your package manager |
| Packages available | AppImage, .deb (Ubuntu/Debian), .rpm (Fedora/RHEL), .pacman (Arch) |
| GPU (optional) | AMD GPU with ROCm for GPU-accelerated OCR |
#100 on February 15, 2026, the .pacman package may fail because of the deprecated http-parser dependency. Use the AppImage instead until this is resolved upstream.
Installation (AppImage)
-
1
Install xdotool
sudo apt install xdotool(Ubuntu/Debian) or equivalent for your distro. -
2
Download and make executable
Download the
.AppImagefile, then run:chmod +x YomiNinja-*.AppImage -
3
Launch
./YomiNinja-*.AppImage— the app will start in your taskbar. If your Linux session uses Wayland, expect mixed results and prefer X11 or XWayland for stable overlay positioning.
macOS
Intel x64 & Apple Silicon (M1/M2/M3)
System Requirements
| Architecture | Intel (x64) and Apple Silicon (arm64) — universal build |
| Permissions | Accessibility access + Screen Recording required |
| OCR engine | Apple Vision Framework (macOS native) + PaddleOCR |
xattr -cr /Applications/YomiNinja.appThis removes the quarantine flag. The app is safe — this is a code-signing limitation, not malware.
Installation
-
1
Download and open the .dmg
Download the
.dmgfile and drag YomiNinja to your Applications folder. -
2
Grant permissions
On first launch, macOS will ask for Accessibility and Screen Recording access. Both are required for the overlay to work.
-
3
Apple Silicon only: remove quarantine
If the app won't open, run
xattr -cr /Applications/YomiNinja.appin Terminal, then try again.
Included
What's included in the download
The app includes the core overlay and dictionary environment, but you still need to import dictionaries and choose the right OCR engine for your content.
PaddleOCR
The default OCR engine. Fast, accurate on standard game text, runs fully offline without a GPU.
Yomitan (pre-installed)
The pop-up dictionary environment is already there. Import JMdict and optional pitch or frequency dictionaries to make it useful on first launch.
10ten Reader (pre-installed)
An alternative Japanese dictionary extension included alongside Yomitan. Useful if you prefer a different popup style or scanning behavior.
Electron runtime
The cross-platform desktop framework. No separate runtime installation needed — it's bundled with YomiNinja.
Chrome Extensions Manager
Install additional Chrome extensions inside YomiNinja's browser context. Compatibility varies, so treat this as optional rather than guaranteed for every extension.
gRPC backend services
Internal communication layer for OCR engine coordination. Runs locally — no cloud dependency for core functions.
Security
Verifying your download
YomiNinja is fully open source (GPL-3.0). You can inspect the source, the build scripts, and the release artifacts on GitHub before running it.
Some antivirus tools may flag OCR or Electron-based apps heuristically. If you see a warning, verify the file hash against the GitHub release assets and compare the behavior against the public source tree rather than relying on generic scanner labels.