Skip to content
Smart Document & Digital Productivity Platform

OCR

Read the text inside images and PDFs and turn it into editable, searchable content.

  1. Add files
  2. Processing
  3. Result

OCR — Extract text

Turn PDFs and images into text you can copy, edit and download.

Add a document to OCR

Drop a PDF or image here. Text extraction starts automatically.

PDF • JPG • PNG • WEBP • TIFF • BMP

Up to 25 MB per file · 10 files at once

No account needed. Your files stay private.

Read the text inside images and PDFs and turn it into editable, searchable content.

What it does

Zegi extracts text from images and PDFs and, where detected, structured fields from invoices, receipts and passports. Confidence scores help with review but do not guarantee that the text or fields are correct.

How it works

  1. Upload an image or PDF.
  2. Zegi reads it with PaddleOCR (primary) and Tesseract (fallback), detecting language automatically.
  3. Review the text and structured fields, then copy or export to Word, Excel or PDF.

Supported formats

Input: images (PNG, JPG/JPEG, WebP) and PDFs. Output: text, structured data, or exports.

What you get

Editable text plus any detected structured data, ready to copy or export.

Important limitations

  • Accuracy depends on the source quality; low-resolution or noisy scans reduce results.
  • Some fields may be missing or misread, even when a confidence score is high. Verify important values against the source.

Privacy & your files

Files are processed on Zegi's own servers. Retention depends on the processing workflow; files may remain for up to 7 days before automatic deletion. Read the Privacy Policy and Data Retention Policy before uploading sensitive documents.

Frequently asked questions

Is OCR free?
Public OCR is free to use; accounts and the API unlock higher limits and additional features. See Pricing.
Which languages are supported?
English and Arabic, including mixed-language documents.
What about invoices and passports?
Zegi can extract structured fields when it recognizes a supported layout. Missing fields may be empty, and detected values can still be wrong. Verify names, dates, identification numbers and amounts against the original.
Is there an API?
Yes — POST /api/v1/ocr, with async jobs and webhooks. See the API documentation.

Related tools

Helpful links