Recognize Text from Image Online — OCR in Browser
Extract text from a photo, scan or screenshot: English, Russian, Chinese, Japanese, more. Runs locally in your browser, no upload. Free.
About text recognition from images
OCR (Optical Character Recognition) extracts printed and handwritten text from photos, scans and screenshots. Drop in an image — the model finds the text, splits it into lines and words, and returns a result you can paste into Word, Google Docs, or send straight to a translator.
Powered by Tesseract.js — an open-source LSTM neural network trained by Google. Supports 16+ languages (English, Russian, Chinese, Japanese, Korean, Arabic, Ukrainian, German, French, Spanish, Italian, Polish, etc.) and an 'Auto' mode with script detection (Cyrillic / Latin / Han / Japanese / Hangul / Arabic).
All processing runs locally in your browser. The image never leaves your device, language packs are downloaded once and cached. No sign-up, no recognition limits. Recognized text can be sent to our translator with one click.
Where it's useful
Extract text from a screenshot
Screenshot of a PDF, slide deck, chat — you need the text but copying isn't allowed? OCR turns the image back into editable text in seconds.
Recognize handwriting or printed documents
Lecture notes, receipts, old books — works great with printed text and handles tidy handwriting reasonably well.
Recognize foreign text and translate it
Menu in a foreign country, Chinese instructions, German signage — recognize and tap 'Translate' to open the text in our translator.
Extract text from phone photos
Snapped a textbook page, schedule, notice? OCR works on mobile, the image stays on your device — private and fast.
FAQ
Which languages are supported?
English, Russian, German, French, Spanish, Italian, Ukrainian, Polish, Portuguese, Czech, Chinese (Simplified), Japanese, Korean, Arabic. Use the 'Russian + English' combo for mixed text. 'Auto' mode picks the language based on detected script.
Is the image uploaded to a server?
No. Recognition is fully client-side — Tesseract.js runs in your browser as WebAssembly. The image never leaves your device, our server logs contain no files or text.
How accurate is it?
For clean printed text, typically 95–99%. Accuracy drops on: handwriting, low resolution (<300 DPI), busy backgrounds, skewed geometry. Tip — shoot perpendicular, in good light, with at least 1500 px on the short edge.
Can it read handwriting?
Tidy handwriting only, and not in every language. Tesseract is trained primarily on printed fonts. For lots of handwriting, consider services with HTR (Handwritten Text Recognition) models.
Can I recognize a PDF?
Yes — use our /pdf/extract-text/ tool, which auto-switches to OCR for scanned PDFs (images inside PDFs). This page handles single images: JPG/PNG/WebP.
How do I translate the recognized text?
After recognition, click 'Translate' — our text translator opens with the text pre-filled. Translation is also local, nothing is uploaded.
How long does recognition take?
First run with a new language: 5–15 seconds (model load, ~5–15 MB). Subsequent runs in the same language: 1–5 seconds for a typical page. Large images (>3000 px) are slower.