📝
Image to Text
Extract text from images with OCR
Last updated:Get the text out of a photo, screenshot or scan so you can copy, edit or search it. Recognition runs in your browser; your images are not uploaded.
Drag & drop your file here
or click to choose
.JPG.JPEG.PNG.WEBP.BMP.TIFF.GIF
Max 50 MB · No registration needed
Your file stays on your device — never uploaded
How to extract text from an image
- 1
Add your images
Drag JPG, PNG, WEBP, BMP, TIFF or GIF files into the box above, or tap to choose them: up to 10 images, 50 MB each.
- 2
Choose the language
Under OCR Language, pick the language of the text, or Auto-detect. For an image in several languages, add up to two more under Additional languages.
- 3
Extract and copy
Press Extract, then Copy Text or Download as .txt.
Good to know
- 81 languages are listed under their own names, such as Deutsch or 日本語. The English page starts with English selected.
- The first run downloads the engine and language data: about 4.5 MB, or 5.5 MB for Chinese, Japanese or Korean. Auto-detect needs about 7.8 MB more. The language data is then kept in your browser.
- Auto-detect checks each file's first image or page and needs a few lines of text; for a short snippet, pick the language yourself. Arabic-script text is read as Persian when letters like پ, چ, ژ or گ appear, otherwise as Arabic. Auto-detect can't be combined with Additional languages.
If something goes wrong
- It says the file type isn't supported
- Convert iPhone HEIC photos with HEIC to JPG first. Save other images as JPG or PNG, then add them again.
- It says the image may be corrupted or unsupported
- Open it in another app, save it as JPG or PNG, and add that copy.
- It says it couldn't download the components
- The engine comes from cdn.jsdelivr.net on first use. Check your connection or pause any blocker, then press Try again.
- The result is mostly nonsense
- Check that the selected language matches the text, or pick it yourself instead of Auto-detect.
- Your file is a PDF
- Use PDF to Text. It copies out text the PDF already contains and only runs recognition on scans.
Frequently Asked Questions
Can it read handwriting?
Only roughly. The engine is built for printed text, so handwriting comes out with many mistakes, joined-up writing most of all.
How do I get a better result?
Use even light, hold the camera straight above the page, crop close to the text and keep it in focus.
How accurate is it?
Sharp screenshots and clean scans come out best; angled photos and small print cause mistakes. Use the Accuracy figure as a guide, and proofread.
Does it keep the formatting?
Line breaks and paragraphs, yes. Fonts, tables and pictures, no: you get plain text.
Are my images uploaded?
No. Only the engine and language files are downloaded; your images are read in your browser and the text stays on your device.