Extract text from a photo or screenshot
Drop a photo or screenshot here, or click to browse
This tool uploads your file — it's deleted automatically in 15 minutes
or paste an image from your clipboard
Photographs of documents are everywhere — a form on a noticeboard, a page of a textbook, a screenshot of a message you cannot copy from. This tool runs optical character recognition on the image and hands back the words as text you can edit, search or paste. English and Hindi are both supported, including pages that mix the two. Unlike most tools here this one needs a server, because a reliable OCR engine is far too heavy to run in a browser on a mid-range phone.
When OCR works and when it does not
Recognition quality is decided almost entirely by the input. Printed text, shot square-on, filling most of the frame, in even light, reads almost perfectly. The same page photographed at an angle under a ceiling light, with a shadow down one side, can come back as gibberish or as nothing at all. If a result looks wrong, retaking the photo is nearly always faster than trying different settings: rest the page flat, avoid your own shadow, and fill the frame with the page rather than the desk around it.
English and Hindi together
Indian forms routinely carry both scripts on the same page — a bilingual header, Hindi labels with English values. Selecting both languages lets the engine consider each, which is more accurate on mixed pages than running either alone. It is slightly slower and very occasionally confuses visually similar characters between scripts, so for a page that is entirely in one language, selecting just that one gives the cleaner result.
Why this one uses a server
Every text, image and most PDF tools here run entirely in your browser, and the pages say so. This one cannot. A usable OCR engine is tens of megabytes of compiled code and needs real processing power per page; the browser builds that exist are slow enough on a mid-range Android to be useless. So the image is uploaded, read, and deleted within fifteen minutes. That trade is stated up front on the upload box rather than buried, because it is a genuine difference in how your file is handled.
Frequently asked questions
- Which languages can it read?
- English and Hindi, separately or together. Pick both for a form or sign that mixes them, which is common on Indian government documents.
- Why did it return nothing?
- OCR needs reasonably sharp, upright printed text. A blurred photo, a strong shadow across the page, text at an angle, or handwriting will usually come back empty. Retaking the photo square-on in even light fixes it more often than any setting.
- Can it read handwriting?
- No, not reliably. The engine is trained on printed type. Neat block capitals sometimes come through, but cursive will not.
- Does it keep the layout?
- Roughly. Line breaks are preserved but columns, tables and headers are flattened into reading order, so a complex form comes back as plain lines rather than a grid.
- Is my image uploaded?
- Yes — this is one of the few tools here that has to be. The engine is too heavy for a browser. The file is deleted automatically within 15 minutes and the text is returned to you, not stored.
- What is the difference between this and OCR a PDF?
- This returns text from a single image. The PDF tool returns a searchable PDF: the original pages with an invisible text layer behind them, so the document still looks like the document.
- Is there a size limit?
- Very large photos are scaled down before reading. A 50-megapixel phone photo of a page carries no more legible text than a smaller one — it just takes far longer — so downscaling costs nothing in accuracy.