YuzuLingo vs transformers_ocr (Ajatt-Tools)
YuzuLingo stays in the browser. You cover burned-in subtitles on the video you are already watching, including YouTube, Bilibili, and other web players. There is no app to set up. Turn on Interact to read the burned-in line on your device, with furigana or pinyin when the language has them. Show that text always, on hover, or while you hold a hotkey.
Transformers OCR is Ajatt-Toolsâ Linux tool, in the repo transformers_ocr. You bind a shortcut, screenshot a manga speech bubble, and it copies the Japanese text. You can also pass an image file. Hold keeps several bubbles, then one recognize copies them together. A config line can send that text to GoldenDict instead of the clipboard. The models stay on the machine, and you can force CPU. It runs on Xorg and Wayland. The sheet says CJK text in images and video frames. The README is Japanese manga OCR from a screenshot. The README says the project is deprecated and to install lancet instead. The owner archived the repo on 8 March 2026. GPL-3.0.
GitHub - Ajatt-Tools/transformers_ocr: đ€ An OCR tool using maim with Transformers.Where each one fits
| Feature | YuzuLingo | Transformers OCR |
|---|---|---|
| Install in the browser, no app to set up | ||
| Cover burned-in subs while a web video plays | ||
| Read that burned-in line while the web video plays, on your device | ||
| Furigana or pinyin on that line | ||
| Show that text always, on hover, or while you hold a hotkey | ||
| OCR Japanese manga text from a screenshot | ||
| Hold several speech bubbles, then copy them together | ||
| OCR stays on the device | ||
| Add the line to an Anki deck |