PyMuPDF Layout is a fast and lightweight layout analysis Python package integrated with PyMuPDF for clean, structured data output from PDF. It's fast, accurate and doesn't need GPUs like vision-based models.
While other tools train machine learning models on rendered page images, PyMuPDF Layout trains Graph Neural Networks directly on PDF internals. This gives us accuracy at 10× the speed utilizing CPU-only resources.
- 📚 Structured data extraction from your documents in Markdown, JSON or TXT format
- 🧐 Advanced document page layout understanding, including semantic markup for titles, headings, headers, footers, tables, images and text styling
- 🔍 Detect and isolate header and footer patterns on each page
PyMuPDF Layout is used by PyMuPDF4LLM to analyze documents and deliver improved results.
PyMuPDF Layout is a component of PyMuPDF4LLM, see the PyMuPDF4LLM documentation page