Image OCR
Extract text from screenshots and photos in your browser with Tesseract OCR — English, Chinese, Japanese, Korean, French, German and Spanish.
Add an image to start.
Drag a rectangle on the image to limit recognition to that area.
Languages
Pick as few as possible — every extra language slows recognition and adds a download. Language data is fetched once from the Tesseract project CDN (2–15 MB each) and then cached by your browser.
Recognised text
How to extract text from an image
- Drop a screenshot or photo into the box. Sharp, straight-on, high-contrast images give by far the best results.
- Select the languages that appear in the image — English only is fastest. On the first run the OCR engine (~5 MB) and each language pack (2–15 MB) are downloaded once and then cached.
- Optionally drag a rectangle over just the paragraph you need, then click Recognise text and copy or download the result.
FAQ
Does my image get uploaded?
No. Recognition runs in a WebAssembly worker inside this tab. The only network traffic is the one-time download of the engine and language data from the Tesseract project's CDN.
Why is the Chinese result inaccurate?
CJK recognition needs larger, cleaner text than Latin script. Zoom in before screenshotting, crop to a single column, and pick Simplified or Traditional — not both.
Can it read handwriting or PDFs?
No. Tesseract is trained on printed text, so handwriting rarely works. For a PDF, export the page as an image first.