PDF OCR
Recognize image-only scanned pages and create a searchable PDF in your browser. Pages that already contain embedded text are kept without re-recognition.
A QUICK WALKTHROUGH
How to use this tool
- Choose one local PDF up to 30 MB and 25 pages.
- Select the language used by the scanned pages and start OCR.
- Download the searchable PDF. The file is processed on this device and is never uploaded.
Scanned pages and existing text
PDF.js checks each page for embedded text. Pages with at least 12 non-whitespace characters keep their existing page content; pages without a text layer are locally rendered and OCRed. This avoids adding a duplicate OCR layer over ordinary digital text pages.
Local OCR and language models
Tesseract.js runs in a browser WebAssembly worker. English, Simplified Chinese, Japanese, and Korean models are served from this site and downloaded on demand. The PDF and recognized text are not sent to a server. Recognition is approximate and should be reviewed.
Limits and output behavior
One PDF up to 30 MB and 25 pages; scanned pages are rendered at 144 DPI, with a 12-megapixel per-page and 80-megapixel total processing limit. OCR pages become raster-backed searchable pages; existing text pages remain as copied page content. Mixed pages that contain both a small text layer and a scan are treated as existing-text pages. Forms, annotations, signatures, outlines, attachments, and document metadata are not guaranteed to survive when a new PDF must be assembled.
GOOD TO KNOW
Common questions
Does this upload my PDF?
No. PDF parsing, page rendering, OCR, and PDF creation run in your browser. Only the selected OCR language model is downloaded from this site.
Will it OCR pages that already have selectable text?
No. Pages with an existing text layer are copied without OCR to avoid duplicated text. Pages with no embedded text are OCRed.
Can it recognize multiple languages in one document?
Choose the main language used by scanned pages. This version processes one OCR language per PDF; recognition errors and mixed-language text are possible.
Does it preserve the original PDF exactly?
No. If scanned pages need OCR, a new PDF is assembled. OCR pages are raster-backed, and interactive features, signatures, outlines, attachments, and metadata are not guaranteed to remain. Keep your original file.