Optical Character Recognition (OCR)

The OCR tool allows you to extract editable text from scanned PDF documents or images.

Supported Input

• Files: PDF documents (.pdf).
• Note: If your PDF has multiple pages, the process may take several minutes per page depending on the density of the text.

Choosing an Engine

Text Lab provides three different AI engines for text extraction:

1. EasyOCR (Default): Good for general purpose text and supports many languages.
2. PaddleOCR: Often performs better on documents with complex layouts or tables.
3. OlmOCR: A specialized pipeline for converting PDFs into clean Markdown.

How to Use

1. Upload PDF: Select your file.
2. Select Engine: Choose one of the engines listed above.
3. Run OCR: Click the button to start processing.
   • Warning: Do not close the tab while the "Running" indicator is active.

Outputs

Once finished, you will see a side-by-side preview of the detected text boxes and the extracted content. You can download: * Extracted Text (.txt) * Structured Data (.jsonl): Useful for developers or data analysis. * Full Package (.zip): Includes text, JSON, and layout images.