What is OCR?
OCR stands for optical character recognition. It is the technology that lets a computer read the letters in a picture. A photo of a page is only a grid of coloured dots to a computer, so you cannot select, copy or search the words in it. OCR finds the lines and letters in the picture and turns them into real text.
This tool uses Tesseract, a well-known open-source OCR engine, running in your browser. It reads printed text in many languages. Handwriting is not supported and is read poorly.
How to get accurate results
OCR is very good on clean, printed text and gets worse as the picture gets worse. These steps make the biggest difference.
- Choose the right language. The engine uses the language to guess letters and words. A wrong choice gives gibberish.
- Use a sharp, well-lit picture. Blur, shadows and glare cut accuracy. Hold the phone steady and avoid your own shadow.
- Photograph the page straight on. Text on a slant or a curved page is harder to read.
- Use enough resolution. Letters should be at least about 20 pixels tall. A screenshot at normal size is usually fine, and the Improve the image option enlarges small pictures for you.
- Crop out the extras. Borders, pictures and decoration around the text can confuse the engine. Crop to the text first.
- Dark text on a light background reads best. Light text on a dark photo is much less reliable.
Supported languages
English, Bengali, Hindi, Arabic, Urdu, Spanish, French, German, Portuguese, Russian, Italian, Turkish, Indonesian, Japanese, Korean and Simplified Chinese. If a page uses two languages, such as a Bengali form with English labels, tick both. Reading is slower with more languages, so choose only the ones you need.
Bengali, Hindi, Arabic and other scripts with joined or marked letters are harder than plain English, so expect to correct more mistakes there, especially in photos. Printed text in a clear font gives the best results.
Your images stay private
Many online OCR sites upload your picture to a server, where it can be stored or used for training. This one does not. The OCR engine and the language files are part of this website and run entirely inside your browser, so ID cards, bank statements, contracts and personal letters never leave your device. You can check this yourself: once the page has loaded, the tool still works with your connection switched off.
The first time you read an image, your browser downloads the engine and the language file you chose from this website, which can take a few seconds. After that, they are kept by your browser and loading is much faster.
What people use OCR for
- Copying text from screenshots, such as error messages, slides and chat messages.
- Digitising paper documents, such as letters, receipts, forms and book pages.
- Translating text in a photo: extract it, then paste it into a translator.
- Searching scanned notes by turning them into text you can search.
- Accessibility. Text in a picture cannot be read by a screen reader, but extracted text can.
Limits
You can add up to 20 images of up to 25 MB each. Reading a page takes a few seconds on a modern computer and longer on an older phone. Handwriting, very stylised fonts, tables, and text in pictures with busy backgrounds are not read reliably, and the layout of columns and tables is not kept. Always check important text, such as numbers and names, against the original.