2403 Resources 0175cdd1

# MinerU Resources

A curated collection of official documentation, community tutorials, and learning materials.

## 📖 Official Documentation

### Primary Sources

- **Official Documentation**: https://opendatalab.github.io/MinerU/
  - Comprehensive guides, API references, and tutorials

- **GitHub Repository**: https://github.com/opendatalab/MinerU
  - Source code, examples, issue tracking
  - 49,600+ stars, actively maintained

- **PyPI Package**: https://pypi.org/project/mineru/
  - Latest version: 2.7.1 (January 2026)

- **Research Paper**: https://arxiv.org/abs/2409.18839
  - "MinerU: An Open-Source Solution for Precise Document Content Extraction"

### Quick Start Guides

- **Installation Guide**: https://opendatalab.github.io/MinerU/quick_start/
- **Quick Usage Guide**: https://opendatalab.github.io/MinerU/usage/quick_usage/
- **FAQ**: https://opendatalab.github.io/MinerU/faq/

### API Documentation

- **Cloud API**: https://mineru.net/apiManage/docs
- **Self-hosted API**: http://127.0.0.1:8000/docs (after running `mineru-api`)

### Related Projects

- **MinerU-HTML (Dripper)**: https://github.com/opendatalab/MinerU-HTML
  - HTML content extraction with AICC dataset

- **PDF-Extract-Kit**: https://github.com/opendatalab/PDF-Extract-Kit
  - Comprehensive PDF extraction toolkit

---

## 🎓 Community Tutorials

### Comprehensive Guides

1. **MinerU Beginner's Guide** by Suke (StableLearn)
   - URL: https://stable-learn.com/en/mineru-tutorial/
   - Published: November 11, 2024
   - **Best for**: Complete walkthrough from installation to advanced features
   - **Covers**: Online/local setup, GPU acceleration, OCR optimization, Python APIs, real-world use cases

2. **Extract Any PDF with MinerU 2.5** by Sonu Sahani
   - URL: https://sonusahani.com/blogs/mineru
   - **Best for**: Real-world testing insights and production use
   - **Covers**: Multi-language testing, hardware requirements, limitations, practical tips

3. **MinerU Document Parsing Tool** by Efficient Coder
   - URL: https://www.xugj520.cn/en/archives/mineru-document-parsing-tool-pdf-markdown-conversion.html
   - **Best for**: Scientific literature extraction workflows

### Technical Deep Dives

4. **From Big Picture to Details: MinerU 2.5**
   - URL: https://aiexpjourney.substack.com/p/from-big-picture-to-details-mineru
   - **Best for**: Understanding the 2.5 architecture and two-stage parsing

5. **Zero-Effort PDF to Markdown** by Hossen
   - URL: https://levelup.gitconnected.com/from-pdf-chaos-to-data-gold-how-mineru-is-revolutionizing-document-intelligence-50180c76e74f
   - Published: December 21, 2024
   - **Best for**: Document intelligence transformation concepts

6. **MinerU: High-Quality PDF Conversion** by Pankaj
   - URL: https://medium.com/@pankaj_pandey/mineru-high-quality-pdf-conversion-for-the-ai-era-c3c2f497936c
   - **Best for**: AI-era use cases for researchers and data scientists

### News and Coverage

7. **MinerU: Open-Source AI Solution** (NeuroHive)
   - URL: https://neurohive.io/en/state-of-the-art/mineru-open-source-ai-document-extraction/
   - **Best for**: Architecture overview and performance metrics

8. **MinerU: Turns Any PDF Into LLM-ready markdown** (Medevel)
   - URL: https://medevel.com/mineru/
   - **Best for**: Features overview and community feedback

9. **MinerU: An Open-Source PDF Data Extraction Tool** (MarkTechPost)
   - URL: https://www.marktechpost.com/2024/10/05/mineru-an-open-source-pdf-data-extraction-tool/
   - **Best for**: Quick technical overview (Editors Pick)

---

## 🎬 Video & Interactive Resources

### Try Online (No Installation)

- **Official MinerU OCR Space**: https://huggingface.co/spaces/opendatalab/MinerU
  - Official Hugging Face demo, try immediately in browser

- **Community Spaces**:
  - https://huggingface.co/spaces/vasilee/MinerU
  - https://huggingface.co/spaces/ApeAITW/MinerU_2.5_Test

### Video Tutorials

**Note**: Dedicated YouTube tutorials are limited. Check these sources:
- OpenDataLab's GitHub for video links
- Search YouTube for "MinerU PDF extraction tutorial"
- Community Discord may have video walkthroughs

---

## 🔄 Comparison Articles

### MinerU vs Alternatives

1. **A Comparative Evaluation of 12 Open-Source PDF Parsing Tools**
   - URL: https://liduos.com/en/ai-develope-tools-series-2-open-source-doucment-parsing.html
   - **Tools**: MinerU, PaddleOCR, Marker, Unstructured, Zerox, Sparrow, Mathpix, and more
   - **MinerU Strengths**: High completion, multi-layout support, 84+ languages, formula/table handling
   - **MinerU Weaknesses**: High resource consumption, no vertical text, struggles with handwritten formulas

2. **Deep Dive into Open Source PDF to Markdown Tools** by Jimmy Song
   - URL: https://jimmysong.io/blog/pdf-to-markdown-open-source-deep-dive/
   - **Tools**: Marker, MinerU, Dolphin, MarkItDown
   - **Key Insight**: "Marker and MinerU are recommended as first choices"
   - **Recommendation**: Marker for book-structured docs, MinerU for open-form docs with complex tables

3. **Which is the Best Model for Document Parsing?** by 302.AI
   - URL: https://medium.com/@302.AI/which-is-the-best-model-for-document-parsing-65405b7d7877
   - **Tools**: Mistral OCR, Markitdown, Jina Reader, MinerU, Doc2X, Dots.OCR
   - **Finding**: "Dots.OCR wins for multilingual mixed-layout PDFs, MinerU for table-heavy biz docs"

4. **Which PDF Parser Should You Use?**
   - URL: https://www.soup.io/which-pdf-parser-should-you-use-comparing-docling-marker-netmind-parsepro-mineru-olmocr
   - **Tools**: Docling, Marker, NetMind ParsePro, MinerU, olmOCR

5. **MinerU Alternatives Catalog**
   - URL: https://alternativeto.net/software/mineru/
   - **Notable Alternatives**: Marker, DocAnalyzer, DeepPDF

6. **MinerU vs Marker Head-to-Head**
   - URL: https://www.aitoolnet.com/compare/mineru-vs-marker

---

## 💬 Community Channels

### Get Help & Connect

- **Discord**: https://discord.com/invite/Tdedn9GTXq
  - OpenDataLab Orals Discord Server (878+ members)
  - Ask questions, share insights, discuss use cases

- **GitHub Discussions**: https://github.com/opendatalab/MinerU/discussions
  - Technical discussions and Q&A

- **GitHub Issues**: https://github.com/opendatalab/MinerU/issues
  - Bug reports, feature requests, troubleshooting

- **DEVONthink Community Thread**:
  - URL: https://discourse.devontechnologies.com/t/the-open-source-project-mineru-is-highly-recommended-as-a-tool-for-pdf-to-markdown/83016
  - Community recommendation and discussion

### Reddit
No dedicated MinerU subreddit; discussions appear in broader tech communities

### Support
- Official FAQ: https://opendatalab.github.io/MinerU/faq/
- DeepWiki AI assistant (mentioned in FAQ)
- WeChat (for direct support, check documentation)

---

## 🛠️ Troubleshooting Resources

### Common Issues & Solutions

**Installation Problems**:
- Missing LibGL on Ubuntu: `sudo apt-get install libgl1-mesa-glx`
- CJK text loss: Install fonts: `sudo apt install fonts-noto-core fonts-noto-cjk && fc-cache -fv`
- Docker recommended for environment compatibility

**Recognition Issues**:
- Complex table errors (rows/columns misidentified)
- Language-specific OCR problems (select correct language for better accuracy)
- Configuration file key-value update errors

**Refer to**:
- Official FAQ: https://opendatalab.github.io/MinerU/faq/
- GitHub Issues for community solutions

---

## 📊 Performance & Benchmarks

### Published Metrics

- **Layout Detection**: 77.6% mAP on academic papers
- **Formula Detection**: 87.7% AP50
- **Formula Recognition**: 0.968 CDM score (comparable to Mathpix)
- **Overall Accuracy**: 82+ (pipeline), 90+ (VLM/hybrid backends)

### Real-World Performance

From Sonu Sahani's testing:
- Multi-page documents processed in seconds
- VRAM usage peaked ~25GB for large documents with vLLM
- Production use: KYC processing reduced from 6-12 min/page to 40-90 seconds

---

## 🚀 Advanced Topics

### Extensions & Integrations

- **Projects Directory**: Community-contributed extensions
  - Multi-GPU processing (LitServe)
  - MCP server integration
  - Asynchronous processing services

### Model Customization
- Custom model paths in `mineru.json`
- Model source selection (HuggingFace vs ModelScope)
- LLM-assisted processing configuration

### Deployment Options
- Docker deployment
- Client-server architecture for distributed processing
- FastAPI REST API
- Gradio web interface

---

## 📚 Related Technologies

### Core Technologies Used by MinerU

**Deep Learning Frameworks**:
- PyTorch, Transformers, ONNX Runtime, Ultralytics YOLO

**Computer Vision**:
- OpenCV, Pillow, scikit-image

**OCR & Document Processing**:
- PaddleOCR (converted to PyTorch), pypdfium2, pdfminer.six

**Inference Acceleration**:
- vLLM, LMDeploy, MLX (Apple Silicon)

**Multimodal Models**:
- Qwen2-VL, UniMERNet, DocLayout-YOLO

---

## 📈 Stay Updated

- **GitHub Releases**: https://github.com/opendatalab/MinerU/releases
- **PyPI Updates**: https://pypi.org/project/mineru/#history
- **Discord Announcements**: Join for news and updates

---

## 🎯 Real-World Use Cases

### Academic Research
- Converting research papers to structured formats
- Extracting formulas (LaTeX) and tables (HTML)
- Building training datasets for LLMs

### Enterprise Applications
- Technical, legal, and business document processing
- Financial report extraction
- Knowledge base construction

### AI/ML Development
- RAG system data preparation
- Custom dataset creation
- Document intelligence systems

### Multilingual Processing
- 109-language OCR support
- International document digitization
- Cross-language knowledge extraction

---

**Have a resource to add?** Submit a PR or share in the Discord community!