OCR extractor: PDF and image to text in 70+ languages

What it does

The OCR extractor converts PDFs and images into clean, machine-readable text while preserving the original layout: tables stay tables, columns stay columns, and reading order survives. It handles more than 70 languages, so a supplier document from any region comes out usable.

Output you can build on

Unlike a plain text dump, layout-preserving output keeps the spatial structure that downstream logic depends on:

• Full text with reading order intact
• Layout structure: tables, columns, headers, and blocks
• Per-region confidence scores for validation and review routing

Where it fits

Use the OCR extractor on its own to make scanned archives searchable, or as the first stage before a specialized extractor: an invoice, purchase order, or bill of lading arrives as an image, OCR makes it text, and the field-level extractor takes it from there.

Input formats

PDFs (native and scanned) and common image formats including JPEG, PNG, BMP, and TIF.