Scanned PDF OCR: Searchable PDF + Text

Turn scanned PDFs and images into searchable PDFs. OCR adds an invisible text layer to every page and keeps the page exactly as it looked, so the file can be searched, copied from and indexed where it is: in your document system, your shared drive or your search index. You also get the text of each page in the dataset.

Most OCR tools give you only the text. This one gives you the document back, searchable, the way scanners in law firms, accounting offices and archives are expected to deliver it, plus the text for your search index or AI pipeline.

What you get

For each file:

• a searchable PDF in the run's key-value store, same pages and images, plus a hidden text layer (optionally PDF/A-2b for long-term archiving);
• the text of each page (on by default), with a flag for pages that already had their own text;
• the number of pages, how many were OCR'd and how many already had text.

Pages that already carry real text (born-digital pages) are kept as they are and charged at a much lower rate, so mixed documents cost less.

Input

Field	What it does
Files	URLs of PDFs or images (JPEG, PNG, TIFF), up to 200 MB each.
Languages	English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Czech, Swedish, Danish, Norwegian, Finnish, Turkish, Romanian, Hungarian. Pick the ones in your documents.
Pages that already have text	Skip (default) keeps them; Redo replaces old OCR text; Force OCRs every page again.
Fix page rotation	Turns sideways and upside-down scans upright.
PDF/A for archiving	Saves PDF/A-2b.
File size	Keep images, optimise losslessly (default) or compress more.
Text per page in the dataset	On by default.
Maximum pages per file	Longer PDFs are refused and not charged (default 500).

{
  "files": ["https://example.com/scans/contract-2019.pdf", "https://example.com/scans/receipt.jpg"],
  "languages": ["eng", "deu"],
  "pdfA": true
}

Output

One dataset item per file:

{
  "url": "https://example.com/scans/contract-2019.pdf",
  "status": "ok",
  "kind": "pdf",
  "pages": 12,
  "ocrPages": 11,
  "textLayerPages": 1,
  "languages": ["eng", "deu"],
  "searchablePdf": {
    "key": "001-contract-2019.pdf",
    "url": "https://api.apify.com/v2/key-value-stores/<store-id>/records/001-contract-2019.pdf",
    "bytes": 2483021
  },
  "pageTexts": [
    { "page": 1, "ocr": true, "text": "SERVICE AGREEMENT\n\nThis agreement is made ..." }
  ],
  "error": null
}

Files that can't be processed (password-protected or digitally signed PDFs, broken downloads, unsupported formats) are listed with the reason and not charged.

Pricing

Pay per event:

• OCR page: each page that goes through OCR.
• Text page: each page that already had its own text and was kept (much cheaper).
• A small start fee per run, set by Apify.

The Maximum cost per run you set in Apify is respected: files not reached are listed as skipped.

How it works

The Actor runs OCRmyPDF with the Tesseract OCR engine, both open source, inside the run. Your files go nowhere else: no third-party OCR service sees them, and no AI service is involved. Output stays in your own Apify storage.

Good to know

• OCR quality depends on the scan: clean 300 DPI scans give the best results; photos at an angle, handwriting and very small print give weaker text.
• Digitally signed PDFs are refused, because changing them would break the signature.
• Keep the language list short: every extra language slows OCR down.

FAQ

Can I use it from my own code? Yes, call it through the Apify API or the Apify clients, and download each PDF from its searchablePdf.url with your Apify token.

Can I run it on a schedule or on new files? Yes, with Apify Schedules or integrations (webhooks, Make, Zapier, n8n).

Support: support+pdfocr@iiclabs.com. Include the run ID if something looks wrong. If it helped you, a rating on this page helps others find it.