Drag an image or click to select
PNG, JPEG, WebP, GIF, BMP
How it works
OCR (optical character recognition) turns the text inside an image —a photo, a screenshot or a scanned document— into editable text that you can copy, search or paste anywhere.
You can choose one or more languages to improve accuracy: setting the correct language helps the engine better interpret accents, special characters and whole words. The first time you use a language its model is downloaded and then stored for next time.
All recognition happens in your browser using WebAssembly: the image is never uploaded to any server. The result includes a confidence estimate so you know how reliable the reading was.
Use cases
- Copy the text from a screenshot when you can't select it.
- Digitize a scanned document, invoice or receipt to edit it.
- Extract a code snippet or an error message from a photo.
- Recover text from an image to translate or search it.
Frequently asked questions
Is my image uploaded to any server?
No. Recognition runs entirely in your browser; the image never leaves your device. Only the language models you choose are downloaded, once.
Why should I choose the text language?
Each language has its own trained model. Selecting the right language noticeably improves accuracy, especially with accents and language-specific characters. If an image mixes languages, you can select several at once.
How do I get better results?
Use sharp, well-lit images with good contrast between text and background. Horizontal, undistorted text works better than angled, blurry or handwritten photos.
What image formats can I use?
You can use the usual formats such as PNG, JPEG, WebP, GIF and BMP. For documents, a screenshot or scan at good resolution gives better results.