How to use Image to Text (OCR)
- Add a photo, a scan, a screenshot (Ctrl+V or Cmd+V) or a PDF, take a picture with your camera, or click the example to see it work.
- Tick the languages in the image. For a photo of a page, leave "Find and straighten the page" on.
- Click "Extract text". The reader downloads once (about 31 MB), then runs on your device.
- Select words on the picture, copy all the text, or download it as .txt, Word, or a searchable PDF.
Why use Image to Text (OCR)
- Finds the page first
- In a photo of a page, the page edges are found by a neural network (on the SmartDoc benchmark of 415 phone-video frames, every corner within 2% of the page), the perspective is removed and the page squared up before reading, as a scanner app does.
- Reads like a scanner app
- Printed English is read by PaddleOCR's PP-OCRv6 model on your device. On a phone photo of a school worksheet, 2% of characters wrong where the previous engine had 46%; on 40 real shop receipts, 8% against 13.5%.
- Tables as tables
- Tables on the page come out as rows and columns you can copy into Excel or Google Sheets as cells, or download as CSV.
- PDFs and searchable PDFs
- A PDF with text gives it directly; a scanned PDF is read page by page, and you can download it back with the text inside, so it is searchable in any PDF app.
- Text on the picture
- The words sit invisibly over the page: press and hold on a phone, or drag on a computer, to select exactly the words you want.
- Arabic and Urdu
- Right-to-left scripts are read with their own engines and displayed correctly; Urdu Nastaliq has a model trained for this site.
About this tool
OCR (optical character recognition) turns pictures of text into editable text. As an image to text converter, it saves retyping invoices, notes, quotes and ID details. Most of the errors people see from OCR on a phone photo are not the recogniser's: the page is at an angle, part of the desk is in the frame, and the lines run downhill. So this page works like a document scanner app first: it finds the four corners of the page, removes the perspective and straightens it, and only then reads the text, at full resolution so small print keeps its pixels.
The reading is done by PaddleOCR's PP-OCRv6 model (Apache-2.
Frequently asked questions
How accurate is it?
Printed text in a reasonable photo is read with about 2% of characters wrong, measured on phone photos of a worksheet and on 40 real receipts; small faint print loses the odd letter. Handwriting is not read reliably. Check names and numbers before relying on them.
Does it support Urdu Nastaliq?
Yes: Urdu has a Nastaliq recogniser trained for this site (see the Urdu OCR page). Choose Urdu alone to use it.
Can I mix English and Arabic in one image?
Yes. Tick both languages. Selecting only the languages that appear gives the best results.
Is my image uploaded?
No. The models are downloaded to your browser once and the image is processed on your device.
How do I get text from a PDF?
Drop the PDF here. Pages that already have text give it straight away; scanned pages are read like photos, and "Download searchable PDF" gives you the same PDF with the text inside.
Does it read tables?
Yes. Tables on the page are shown as rows and columns under the text, with Copy (pastes into Excel or Google Sheets as cells) and a CSV download. Merged headers and very dense tables can need a correction.
Can I copy text from a screenshot?
Yes. Paste the screenshot with Ctrl+V (or Cmd+V on a Mac), or choose it from your files, and the text appears ready to copy or download as a TXT file.
Is this an image to text extractor?
Yes. It extracts the text from a photo, screenshot or scan so you can copy it, edit it or save it.
Can I take a picture and convert it to text?
Yes. On a phone, tap "Take a photo" and shoot the page; the picture opens here, ready for "Extract text". On a computer, "Use your phone's camera" gives a code to scan, so you can photograph a book page or a letter with the phone and copy its text on the computer. A webcam works too, for a page held up to it. Hold the camera square to the page and let the text fill the frame.
Image se text copy kaise kare, text kaise nikale? (How do I copy text from an image on my phone?)
Photo ya screenshot daalein aur "Extract text" dabayein; text copy karna ho to "Copy text" dabayein. Add the photo or screenshot and let it read; then press and hold the words on the picture to select them, or tap "Copy text" for the whole page. The AI that reads the text runs on your device, so the image is never uploaded.
Does it work on iPhone and Android?
Yes. To copy text from an image on an iPhone or convert an image to text on Android, take the photo or choose a screenshot from your gallery, and the text is read on the phone, in the mobile browser with no app (the reader downloads once, its size shown first). Copy it, or save it as a text or Word file. On a computer, "Use your phone's camera" shows a code to scan, so you can photograph the page with your phone and copy its text on the computer.
Built and maintained by Saad Naeem
AI engineer, Doha · Updated