Image to Text (OCR)

What Image to Text (OCR) helps you do

Turn a photo, screenshot, or scan into editable text without sending the image anywhere. Optical character recognition runs inside the browser through Tesseract, so a receipt, a slide, a whiteboard photo, or a screenshot of an error message becomes text you can copy into a document, issue tracker, or email. It is built for one image at a time, which keeps the result easy to check.

Image recognition runs in a Web Worker inside your browser using Tesseract.js. Your image is never uploaded; only the language data file is downloaded once and cached by the browser.

How it works

The image is passed to Tesseract, a recognition engine compiled to WebAssembly and run in a Web Worker. The selected language data trains the engine for that script. The worker returns the recognised text, which is displayed for review; the image is never sent to a server, and the engine releases the worker after the run.

How to use Image to Text (OCR)

1. Choose the language used in the image, then drop the image or browse to select it.
2. Select Extract text and watch the progress bar while the language data loads and recognition runs.
3. Review the recognised text and correct misread characters.
4. Copy the text or download it as a .txt file.

When to use it

• Copy text out of a screenshot instead of retyping it.
• Digitise a printed page, receipt, or label captured with a phone camera.
• Make notes from a whiteboard or presentation photo searchable.
• Extract code or an error message from a screenshot for a bug report.
• Read text from an image that has no accessible text layer.
• Turn a printed form into editable text for a document.

When not to use it

• Do not expect reliable handwriting recognition; Tesseract is built for print.
• Do not use one image for a multi-page document; use the PDF OCR tool.
• Do not rely on the output for numbers, names or legal text without proofreading.

Example

A screenshot of a database error message is dropped in with English selected. The tool returns the stack trace text, which is pasted into a bug report without retyping it. A phone photo of a receipt is also readable, but the total is checked against the image because OCR can confuse 8 and 3.

What you can do with Image to Text (OCR)

• Recognise printed text in the languages offered by the selector.
• Read photos, screenshots and scanned pages.
• Show a progress bar while language data loads and recognition runs.
• Copy the recognised text with one click.
• Download the result as a .txt file.
• Keep the image and the text on your device.

Example: copy text from a screenshot

Choose the language that matches the image, then drop the screenshot. Selecting the wrong language is the most common reason for garbled output.

Run recognition and read the result against the image, especially numbers, names and punctuation. OCR often confuses 0/O, 1/l and similar pairs.

Correct the text in place, then copy it or download the .txt. For a multi-page scan, use the PDF OCR tool instead.

What to know before you start

• Printed text works best. Handwriting and decorative fonts are recognised unreliably.
• Low resolution, skew, glare, and low contrast increase errors; 300 DPI scans are ideal.
• The first run in each language downloads about 10 to 15 MB of language data.
• Each recognition covers one image; use PDF OCR for multi-page documents.
• Proofread the output before relying on it for numbers, names, or legal text.

Common mistakes

• Leaving the language on the default when the image is in another language.
• Using a low-resolution or skewed photo and expecting clean text.
• Trusting numbers and codes without checking them against the image.
• Running OCR on a multi-page PDF instead of using the PDF tool.
• Assuming handwriting will be recognised accurately.

Troubleshooting

If the text is garbled, check the selected language first, then the image quality. Skew, glare and low contrast all increase errors.

If the first run is slow, the language data is downloading. Later runs reuse the browser cache.

If recognition fails on a very large image, resize it or crop the text region before running OCR.

If the text is mostly punctuation, the image may be too small or the contrast too low. Try a sharper crop.

If a language is missing from the selector, the tool does not ship that data file.

Related standards and documentation

• Tesseract is the open-source OCR engine used by the browser build.
• The output is plain text; no hOCR or searchable PDF layer is produced.