Image to Text (OCR)

How to use it

1. Choose the language of the text.
2. Add an image.
3. Review the result, then copy it.

Good to know

Choosing the right language makes a big difference. Language data is downloaded the first time you use it.

Clear, straight, well-lit printed text gives the best results. Handwriting is rarely read correctly.

How to get an accurate result

• Choose the matching language before adding the image. Reading French with the English setting will damage the accents and some words.
• Make sure the letters are large enough. For scans, aim for 300 DPI, and for photos, move closer rather than zooming in.
• Keep lines level. Slanted text and pages photographed at an angle cause many of the errors.
• Crop out everything except the text, including table edges, logos and neighbouring pages.
• Dark text on a plain, light background gives the best results.

Languages available

English, Spanish, German, French, Portuguese, Italian, Polish, Ukrainian, Russian and Turkish. The recognition engine is Tesseract, running in your browser. The data for each language is downloaded from jsDelivr the first time you use it, and your image is not part of that request.

What to check in the result

Limits of the tool

Handwriting is rarely read correctly, since the engine is trained on printed text. Multi-column pages and tables can come out in the wrong order, so crop each column and process it separately. The tool handles one image at a time. For a scanned PDF, convert it with PDF to JPG first.

Questions