OCR Text Extractor
Extract text from images and scans in 16 languages with Tesseract OCR, running in your browser.
Why use this tool?
Extract text from photos, screenshots and scanned documents with Tesseract OCR, in 16 languages including English, Bengali, Hindi, Arabic and Chinese. The image is processed in your browser; only the language model is downloaded the first time.
How to use it
- Choose the language of the text.
- Upload a clear, well-lit image.
- Copy the recognized text and fix any small errors.
What OCR does
Optical character recognition turns a picture of text into text you can edit, search and copy. This tool uses Tesseract, a widely used open-source OCR engine originally developed at HP and later maintained with Google's support, compiled to run inside your browser.
Your image is never uploaded. The first time you use a language, its trained model (a few megabytes) is downloaded and cached by your browser; after that, recognition runs entirely on your device. Sixteen languages are available, including English, Bengali, Hindi, Arabic and Chinese.
Getting accurate results
OCR accuracy depends mostly on the image. Text should be in focus, evenly lit and reasonably large: aim for capital letters at least 20-30 pixels tall. Photograph pages straight on rather than at an angle, avoid shadows from your phone, and crop away anything that isn't text.
Screenshots and clean scans work best, often with near-perfect results. Low-contrast text (grey on grey), decorative fonts, text over photos and handwriting are much harder; handwriting in particular is usually recognised poorly by Tesseract.
Always pick the language the text is written in. Recognising Hindi text with the English model, for example, produces nonsense, because each model only knows its own alphabet and common words.
After recognition
Proofread the output before using it. Typical errors are swapped look-alike characters (0 and O, 1 and l, rn and m), lost formatting in tables, and line breaks in the middle of sentences where the original had a narrow column.
Multi-column layouts such as newspapers can come out in the wrong order. Cropping each column into its own image and running them separately fixes this.
Frequently asked questions
Is my image uploaded anywhere?
No. Only the language model is downloaded the first time; the image itself is processed in your browser.
Can it read handwriting?
Only very neat, print-style handwriting, and even then with errors. Tesseract is designed for printed text.
Can I extract text from a PDF?
If the PDF already contains text, it's better to copy it directly. For a scanned PDF, convert the page to an image first (for example with the PDF to JPG tool), then run OCR on it.