Image to Text Converter (OCR)
Extract text from photos, screenshots and scans — free OCR that runs in your browser.
Text reading runs in your browser with Tesseract. The first use downloads the reading engine (about 3 MB), then it's cached. Best results: clear, straight, well-lit text.
Was this tool helpful?
Thanks for the feedback!
This OCR tool reads the printed text in a photo, screenshot or scan and gives it back as editable text you can copy. It runs Tesseract, a well-established open-source text recognition engine, directly in your browser, so a photographed contract, ID card or bank letter is never uploaded to a server.
How to extract text from an image
- Choose an image with Upload image. JPG, PNG, WEBP and BMP files all work.
- Click Extract Text. On first use the browser downloads the OCR engine and English language data, which needs an internet connection and takes a moment; later runs use the cached copy.
- Watch the progress bar as the engine works through the image.
- Review the result in the Extracted text box, correct anything you need to directly in the box, then press Copy text.
A PDF can't be uploaded directly; convert scanned pages to images first with PDF to JPG. iPhone HEIC photos are not among the formats the engine reads, so run those through HEIC to JPG first.
What it reads well, and what it struggles with
| Reads well | Struggles with |
|---|---|
| Screenshots of apps, websites and error messages | Handwriting, especially joined-up |
| Printed pages, notices and letters | Decorative or very stylised fonts |
| Straight, sharp photos of documents | Text on busy or patterned backgrounds |
| Dark text on a light background | Hindi and other non-Latin scripts |
The tool loads only the English model, so it recognises the Latin alphabet, digits and common punctuation. Devanagari and other Indian scripts will not come out correctly, and accented letters may be misread.
Getting cleaner results
- Straighten and crop. Rotate the image so lines run horizontally, and crop away everything that isn't text, such as table edges, fingers or backgrounds.
- Resolution matters. Tesseract's own guidance favours scans at around 300 DPI. When photographing, fill the frame with the page instead of zooming a small area later.
- Light evenly. Shadows across a page, glare on glossy paper and low contrast are the most common causes of garbled output.
- One column at a time. Multi-column layouts, like newspapers, can be read across the columns. Crop each column separately for text in the right order.
Tidying the extracted text
OCR keeps a line break at the end of every printed line, so pasted paragraphs look choppy. The remove line breaks tool joins them back into flowing text. Tables come out as plain lines of words without columns, so rebuild them in a spreadsheet if needed.
Always proofread numbers, names and amounts. Classic OCR confusions include 0 and O, 1, l and I, 5 and S, and "rn" read as "m". A single wrong digit in an account number or invoice total matters far more than a typo in a paragraph. If the box says no text was detected, try a larger, sharper and better-lit version of the image.