Image to Text (OCR)
Copy text out of a photo, screenshot or scanned page. Recognition runs on your device, so documents stay private.
Your file will appear here.
Getting the best results
- Use a sharp, well-lit photo taken straight on. Shadows and angles reduce accuracy.
- Screenshots and scans work best; handwriting is usually not recognised reliably.
- Pick the right language. Mixed-language documents work best with the main language selected.
- Crop away everything that isn’t text first with the Crop Image tool.
The first run for each language downloads its model (a few MB). After that it’s cached and much faster.
How to use the Image to Text (OCR): step by step
- Choose a photo, screenshot or scan containing text.
- Select the language of the text.
- Click Extract text.
- Copy the result and check it before using.
A real example
A colleague sends a screenshot of a price list. Extract the text, copy it into a spreadsheet, and you don’t have to retype 40 lines.
Tips for better results
- Straight, sharp, well-lit images give the best accuracy.
- Screenshots work better than photos of screens.
- Handwriting is usually not recognised reliably.
More questions
Which languages are supported?
English, Urdu, Arabic, Hindi, French, Spanish, German, Portuguese, Italian, Turkish, Indonesian and Chinese (Simplified).
Why are some words wrong?
Shadows, angles or low resolution confuse recognition. Retake the photo straight on, or crop to just the text.
Frequently asked questions
Are my images uploaded?
No. Images are processed by your browser on your own device, so private photos never leave it.
How accurate is it?
For clear printed text, usually over 95%. Accuracy drops with blur, low light, decorative fonts or handwriting, so always proofread.
Does it work with Urdu and Arabic?
Yes. Choose Urdu or Arabic in the language list. Right-to-left scripts are supported, though printed text gives far better results than handwriting.