Skip to main content
PDFTulip

OCR a scanned PDF or image

or drag and drop files here

PDF, JPG, PNG, WebP · up to 50 MB per file

Privacy: this tool runs in your browser. Your file is not uploaded to our servers. How we handle files

How to use OCR PDF

Turn pictures of text into real text. You get the words to copy or save, and a PDF that looks like your scan but can be searched and selected. Recognition happens on your device, which takes a few seconds per page.

  1. Choose a scanned PDF, or one or more photos (JPG, PNG or WebP). Photos become pages in the order of the list.
  2. Pick the language of the text. The first time, a language file of 1 to 2 MB is downloaded from this site and kept by your browser.
  3. Select Recognise text and wait. You can cancel at any time.
  4. Copy the text, or download the searchable PDF or a .txt file.

Supported formats and limits

  • Input: PDF, JPG, PNG, WebP, up to 50 MB per file.
  • Up to 50 pages per run in total (PDF pages plus pictures), and up to 30 files.
  • Everything runs on your device, so time depends on your device: roughly a few seconds per page on a laptop, longer on a phone. Large runs can take several minutes.
  • Accuracy depends on the scan. Clear, straight, printed text at 200 DPI or better reads well. Handwriting, small or blurry type, curved pages, stamps and busy backgrounds read badly. Always check names, numbers and anything important.
  • Pages are read as one block of text. Tables and several columns can come out in a wrong order, and tables are not rebuilt as tables.
  • Pages are not turned automatically: a sideways or upside-down page needs Rotate PDF first.
  • The searchable PDF is rebuilt from page images at 200 DPI, so it replaces any text, vector graphics, links and form fields the original had and its size can differ from the original. Keep your original file.
  • Languages: English, Spanish, French, German and Hindi. The search layer of the PDF holds Western European text only, so Hindi is given as text only, not as a searchable PDF.
  • Password-protected PDFs are not accepted here. The tool reports them and stops; remove the password first with Unlock PDF (you must know it), then use the unlocked copy.

Frequently asked questions

What is the difference between this and PDF to Text?

PDF to Text reads text that is already stored in a PDF, which is instant and exact. OCR looks at the picture of each page and guesses the letters, which is slower and can contain mistakes. Use OCR only when the PDF has no selectable text.

Are my files uploaded for recognition?

No. The recognition program, its language files and your pages all stay in your browser. The only things downloaded are the program and the language file, from this website.

How accurate is it?

On clean printed text it is usually good, but it is never perfect and this page cannot promise a figure. The confidence shown after a run is the engine's own estimate, not a measured accuracy. Read the result before you rely on it.

Why is the searchable PDF larger or different from my original?

It is made from a picture of each page with an invisible layer of text on top. That keeps the look of a scan but replaces the original page content, so the size changes and original links or form fields are gone.

Can it read handwriting?

Not reliably. It is built for printed text, and handwriting usually comes out as wrong or missing words.

Created and maintained by Jasmin Gohil. Last reviewed .

We use essential storage, and with your permission Google Analytics to count visits. We never send what you enter or upload. Privacy policy