OCR, Image to Text
Extract text from images, scans and screenshots with OCR, then copy or edit it. Free, works in many languages, and runs in your browser without upload.
Drop, click, or paste (Ctrl+V)
PNG, JPG, WebP, GIF - any image with text
Pull text out of images, privately
Drop in a photo of a document, a screenshot, a receipt or a book page and get selectable, copyable text. Recognition runs entirely in your browser using Tesseract — the image is never uploaded, which makes this safe for contracts, ID documents, medical paperwork and anything else you would hesitate to send to an unknown server.
Accuracy depends mostly on the input: sharp focus, even lighting and reasonably straight text recognise dramatically better than skewed phone photos. The tool pre-processes images before recognition — upscaling small text and boosting contrast — which rescues many marginal scans.
Expect near-perfect results on printed text and screenshots, and rougher results on handwriting, which remains genuinely hard for OCR. For multi-page documents, process pages one at a time and paste as you go.
Frequently asked questions
Which languages does it recognise?
English, Spanish, French, German and Simplified Chinese. Pick the language before pressing Extract Text; the matching language data is loaded on demand.
Is my image uploaded?
No. The Tesseract engine and language data are downloaded from a CDN, but the image is processed in your browser and never sent anywhere.
Can it read handwriting?
Only roughly. Tesseract is trained on printed text, so neat block capitals may come through while cursive usually does not. Expect to correct the output.
Can it read a PDF?
No, it takes images only: PNG, JPG, WebP or GIF, dropped, chosen from your files or pasted with Ctrl+V. Screenshot the PDF page or export it as an image first.
What does the confidence percentage mean?
It is Tesseract's own average certainty for the recognised text. Below 70% the tool flags the result; a sharper image and the right Layout setting (single block, single line or scattered text) usually raise it.