Image to text (OCR)

How to pull text out of an image or a scan

Upload a photo, a screenshot or a scanned PDF and start recognition. The text appears in a box, ready to copy or download as TXT. You can also fix it up right there.

Recognition runs on the PaddleOCR model on our server, noticeably more accurate on photos of documents than in-browser engines. The document is processed in memory and deleted the moment the result is sent back; it is never stored.

• Handles English and Czech at once, accents included, no language picker.
• For a scanned PDF it renders the pages and recognises them one by one, up to twenty.
• A multi-page PDF counts as a single use of the daily limit.

How to get a better result

Recognition lives and dies by the source image. Shoot the page square on from above, not at an angle, and watch the focus. A shadow across half the page, or a handheld shot in dim light, turns the result into an unreadable mess.

Frequently asked questions

Does it read handwriting?

No. Tesseract is built for printed text. Handwritten notes come out right only by accident.

Why does the text have typos?

When confidence is low, a sharper scan, more contrast or the right language choice all help. Text with accents recognised using the English model comes out stripped of them.

Are the layout and tables preserved?

No, the output is plain text. Columns and tables collapse into lines, so they usually need rebuilding by hand.

Does it work on an ordinary PDF with text?

It does, but there is no point. If the PDF already contains text (you can search it), use the PDF to Word tool, the result will be more accurate.