Local processing · No uploads
Image to Text
Use Ctrl+V / ⌘V to paste a screenshot while this page is focused.
Some pages produced no recognized text.
One image, up to 20 MiB, 16 million pixels and 8,000 pixels per side. Only the first frame of animated images is processed.
OCR starts only when you select Recognize. Engine and language models download from this site (English model ≈3 MiB; Chinese ≈2 MiB, plus the engine). Model files may be cached in this browser. Your images, PDFs and recognized text are never uploaded or saved automatically.
How to use
Choose an image and its language, then recognize. You can cancel while resources load or during recognition. Copy the result or download UTF-8 TXT.
Recognition quality
Use sharp, upright images with readable printed text. OCR may miss or misread characters; check names, numbers and punctuation. Handwriting and complex tables are not guaranteed. This is a local OCR tool, not an AI document analysis service.