Does the PDF look different after OCR?

No. The original page content is kept as is; the recognised text is added invisibly at the matching positions and only shows up when you search, select or copy.

Can I search and copy Chinese text afterwards?

Yes. The invisible layer is written as Unicode with a ToUnicode map, so Chinese can be searched and copied in Adobe Acrobat, Chrome, Edge, macOS Preview and other readers. Selection is aligned per line; the box around each individual character is approximate.

How accurate is the recognition?

Very good on clear printed documents. For tiny text or blurry scans, switch to 300 DPI for better accuracy; handwriting, decorative fonts and heavily skewed pages work less well.

Are my files uploaded?

No. Once the model is downloaded to your browser, rendering, recognition and writing the text layer all happen locally — suitable for contracts, IDs and other sensitive scans.

👁️

👁️PDF OCR - Make Scanned PDFs Searchable and Copyable Online

Scanned contracts, invoices, books and archives are image-only PDFs: you can’t search them or copy a word out of them. This tool renders each page to an image in your browser, runs a PaddleOCR model for English and Chinese to find every line of text and where it sits on the page, and then adds an invisible text layer (PDF rendering mode 3) on top of the original page. The page looks exactly as before, but Ctrl+F finds your keywords, you can select and copy text, and document systems and search engines can index it. Chinese doesn’t require embedding a multi-megabyte font: the invisible layer uses a glyph-less font of a few hundred bytes plus a Unicode map — the same technique as Tesseract and OCRmyPDF — so Chinese, English and digits all search and copy correctly while the file barely grows. Recognise only certain pages, skip pages that already have text, cancel at any time, and download the recognised text as TXT or copy it in one click. The first run downloads a ~15MB model that is cached afterwards, and your file never leaves your computer.

Free · No sign-up · Runs in your browserLast updated: 2026-10-26

Loading tool...

How to use PDF OCR

1Upload a scanned PDF; the recognition model downloads automatically on first use
2Choose the resolution and page range; pages that already contain text are skipped by default
3Click Start OCR, then download the searchable PDF or download/copy the recognised text

FAQ

Does the PDF look different after OCR?: No. The original page content is kept as is; the recognised text is added invisibly at the matching positions and only shows up when you search, select or copy.
Can I search and copy Chinese text afterwards?: Yes. The invisible layer is written as Unicode with a ToUnicode map, so Chinese can be searched and copied in Adobe Acrobat, Chrome, Edge, macOS Preview and other readers. Selection is aligned per line; the box around each individual character is approximate.
How accurate is the recognition?: Very good on clear printed documents. For tiny text or blurry scans, switch to 300 DPI for better accuracy; handwriting, decorative fonts and heavily skewed pages work less well.
Are my files uploaded?: No. Once the model is downloaded to your browser, rendering, recognition and writing the text layer all happen locally — suitable for contracts, IDs and other sensitive scans.