📄 PDF Zoo Wiki

Welcome to the PDF Zoo - a comprehensive collection of tools, libraries, and models for PDF processing, OCR, and document understanding. From traditional OCR engines to cutting-edge LLM-powered solutions, find the right tool for your document processing needs.

🧩 The PDF Parsing Challenge

PDF parsing has evolved from simple OCR to sophisticated layout understanding and vision-language models. Learn about the history and ongoing challenges of document processing and why different tools exist for different needs.

⭐ My Faves

• olmocr - Currently one of the most accurate OCR tools available, leveraging cutting-edge vision-language models
• PyMuPDF - Excellent documentation, cost-effective to run, and reliable for most PDF processing tasks
• surya - Great performance with bounding boxes and preserved reading order for layout-aware extraction
• s2orc-doc2json - Excellent for parsing and structuring academic papers down to the section level, with robust citation extraction
• MinerU - More complex but impressive at intelligently detecting and separating content from boilerplate text

🚀 Quick Start

Explore our curated collection organized by type and capability in the sidebar, or jump straight to these key categories:

📚 OCR Solutions

• Traditional OCR - Time-tested engines like tesseract
• Model-Based OCR - Deep learning tools: easyOCR, PaddleOCR, textra
• LLM-Based OCR - Vision-language models: olmocr, GOT-ocr2, florence
• Layout Analysis - Structure-aware: surya, grobid, publaynet

🔧 Processing Tools

• OCR Toolkits - PyMuPDF, docling, OCRmyPDF
• PDF to Markdown - marker, s2orc-doc2json, nougat
• Text Extraction - pdfplumber, pdfminer.six, pypdf2
• Text Structuring - NuExtract, langextract

☁️ Cloud Services

• Enterprise APIs - Azure Document Intelligence, AWS Textract, Google Document AI

📋 Entry Format

Each tool entry follows this standardized format:

### [Tool Name](repository-url)
tags: `#category`, `#feature1`, `#feature2`  
deps: [dependency1](link), [dependency2](link)  
inst: `organization/company`  
paper: https://arxiv.org/abs/paper-id  
date: Month Year  
live: https://demo-url.com  
models: [model_name](link)  
limitations: Brief description of constraints or downsides   
> Brief description of capabilities and key strengths

<img src="workflow-diagram.png" alt="Tool workflow" width="600">

🏷️ Tag System (TAGxonomy)

Our classification system helps you find tools by type and functionality:

Primary Categories

Tag	Description
#trad	Traditional/classic OCR approaches
#model-based	Modern neural network-based OCR
#llm	Large language model-based OCR
#toolkit	Multi-purpose libraries with various features
#ocr-free	Transformer models that bypass traditional OCR
#cloud	Enterprise cloud-based document processing services

Functionality Tags

Tag	Description
#readingOrder	Text reading sequence determination
#layoutAnalysis	Document layout understanding
#pdf2markdown	PDF to markdown conversion
#layout	General layout processing
#multiModal	Multiple input types support
#PDF-wrangling	PDF manipulation and processing
#tableExtraction	Table detection and extraction
#formulaOCR	Mathematical formula recognition
#handwriting	Handwritten text recognition
#multilingual	Multi-language support
#structuring	Unstructured to structured text
#gpu	GPU acceleration support
#addTextLayer	Adds searchable text layers to PDFs

🤝 Contributing

Found a tool we're missing? Want to suggest improvements? This wiki is a living document that grows with the community's contributions.