OCR for Scanned Documents
Recognize English text from images or scanned PDF pages.
English OCR runs locally. Up to 10 PDF pages per job. Recognition can make mistakes, especially with handwriting or complex layouts; check the text.
How to use OCR for Scanned Documents
- Choose a clear, upright scan. For a PDF, choose a short page range such as 1-3; unlock encrypted PDFs first.
- Start OCR and wait for the English model to load. Cancel stops recognition; reducing the page count helps on smaller devices.
- Compare the recognized text with the original, especially names, dates and amounts. Download the plain-text result before leaving the page.
Continue your workflow
Extract data from a scan — Turn scanned information into editable text and review it.
What to expect
Recognize English text in scanned PDFs, PNG, JPEG or WebP. Inputs are limited to 10 MB, image inputs to 12 megapixels, and PDF selections to 10 pages. The complete job has a two-minute limit. OCR is local; English model assets load from this site. Handwriting, low contrast, rotated scans and complex tables may produce errors.
Local processing
Your files and inputs are processed in your browser and are not uploaded. Download or copy results before leaving. Normal site and library requests still reach the hosting service.
Frequently asked questions
Why are columns or numbers incorrect?
OCR recognizes shapes, not document meaning. Use a sharper, upright scan with good contrast. Multi-column reading order and similar characters such as O/0 require manual review.
Does OCR preserve the document layout?
The result is plain text, with page labels for PDFs. It does not recreate tables, formatting or a searchable PDF. PDF to Text is a faster choice when the original already contains selectable text.
What if loading fails or the time limit is reached?
Reload to retry asset loading, then try one page or a smaller image. Cancellation during model startup discards a worker that finishes starting later; browser resource cleanup still needs verification.
Are my files uploaded?
Your files are processed locally in your browser and are not uploaded. Download the result before closing the page; normal website and processing-library requests still reach the hosting service.