Skip to content

Optical Character Recognition (OCR)

The OCR tool allows you to extract editable text, tables, and structured data from scanned PDF documents or raw images.

Supported Input

  • Documents: PDF (.pdf). Note: If your PDF has multiple pages, the process may take several minutes per page depending on the density of the text.
  • Images: PNG, JPG, JPEG, BMP, TIFF
  • Archives: ZIP (.zip) (for batch processing multiple documents or images at once).

Choosing an Engine

Text Lab provides four different AI engines for text extraction, each with its own strengths:

  1. EasyOCR: Good for general-purpose text and supports a wide variety of languages.
  2. PaddleOCR: Often performs better on documents with complex layouts or multi-column text.
  3. OlmOCR: A specialized pipeline optimized for converting scientific PDFs into clean Markdown.
  4. GLM-OCR: A state-of-the-art large vision model. It is exceptional for highly complex layouts and allows you to specifically target what to extract (e.g., General Text, specific Tables, or Figures).

How to Use

Single Document OCR

  1. Select Workflow: Choose "Single Document OCR".
  2. Upload File: Select your PDF or image.
  3. Select Engine: Choose one of the engines listed above.
  4. If using EasyOCR/PaddleOCR: Select the language of the document.
  5. If using GLM-OCR: Select the extraction mode (Text, Table, or Figure).
  6. Run OCR: Click the button to start processing.
  7. Warning: Do not close the tab while the "Running" indicator is active.

Batch Processing (Multiple Files)

If you have a large dataset of documents or images (e.g., a folder of 50 scanned PDFs), you can process them all in one go:

  1. Select Workflow: Choose "Batch OCR (ZIP)".
  2. Upload File: Compress all your PDFs or images into a single .zip archive on your computer and upload it.
  3. Select Engine: Choose the OCR engine (and language/mode if applicable). These settings will be applied to all files in the ZIP.
  4. Run Batch OCR: Click Start. The AI models are loaded into the GPU only once, and the system loops through your entire archive at maximum speed.

Outputs

For single documents, you will see a side-by-side preview of the document and the extracted content. You can download:

  • Extracted Text (.txt): The raw text or Markdown.
  • Table Data (.csv): If the AI detects a structured table, you can download it directly as a CSV for Excel/Python.
  • Structured Data (.jsonl): Useful for developers or bulk data analysis.
  • Full Package (.zip): Includes all text, JSON, and layout images in one convenient package.

(Note on Batch Processing: When running a Batch OCR, you will receive a single ZIP download. This ZIP perfectly recreates your original uploaded folder structure, with each document's extracted .txt and .json data neatly organized into individual folders!)

Data Privacy & Security

Text Lab is designed to handle highly sensitive, confidential, and proprietary documents (including unredacted PDFs, medical records, or unpublished manuscripts). We utilize a strict "Local-Only, Self-Cleaning" architecture to ensure your documents remain secure.

Here is exactly what happens to your data when you use the OCR tool:

  • 100% Local Processing: Your documents are never sent to external cloud services or APIs (like Adobe, Google Cloud Vision, or AWS). All text extraction is performed entirely on the University of Bern's secure UBELIX high-performance computing nodes.
  • Isolated User Workspaces: While simple text is processed in memory, advanced AI vision models (like OlmOCR) and Batch ZIP processing require reading files from a hard drive. To accommodate this, Text Lab generates a unique, temporary workspace located strictly within your private University home directory ($HOME/ondemand_text_lab_ocr_jobs). Other users on the cluster cannot access this space.
  • Instant Auto-Deletion (Self-Cleaning): The exact moment the AI finishes extracting the text (or if the process encounters an error), the application runs an aggressive shutil.rmtree() command. This guarantees that the entire temporary workspace—including your original documents, intermediate images, and raw data files—is instantly and permanently deleted from the hard drive.
  • Ephemeral Results: The final extracted text and tables presented on your screen are stored strictly in your browser's volatile memory (st.session_state). When you close the tab, refresh the page, or your HPC job ends, all traces of the document and its extracted text are destroyed by Python's garbage collector.
  • No AI Training: The vision models only perform inference (looking at the image to extract text). They do not learn from your documents, and your data is never used to train or improve the AI.