- 2024/11/22 0.10.0 released. Introducing hybrid OCR text extraction capabilities,
- Significantly improved parsing performance in complex text distribution scenarios such as dense formulas, irregular span regions, and text represented by images.
- Combines the dual advantages of accurate content extraction and faster speed in text mode, and more precise span/line region recognition in OCR mode.
- 2024/11/15 0.9.3 released. Integrated RapidTable for table recognition, improving single-table parsing speed by more than 10 times, with higher accuracy and lower GPU memory usage.
- 2024/11/06 0.9.2 released. Integrated the StructTable-InternVL2-1B model for table recognition functionality.
- 2024/10/31 0.9.0 released. This is a major new version with extensive code refactoring, addressing numerous issues, improving performance, reducing hardware requirements, and enhancing usability:
- Refactored the sorting module code to use layoutreader for reading order sorting, ensuring high accuracy in various layouts.
- Refactored the paragraph concatenation module to achieve good results in cross-column, cross-page, cross-figure, and cross-table scenarios.
- Refactored the list and table of contents recognition functions, significantly improving the accuracy of list blocks and table of contents blocks, as well as the parsing of corresponding text paragraphs.
- Refactored the matching logic for figures, tables, and descriptive text, greatly enhancing the accuracy of matching captions and footnotes to figures and tables, and reducing the loss rate of descriptive text to near zero.
- Added multi-language support for OCR, supporting detection and recognition of 84 languages.For the list of supported languages, see OCR Language Support List.
- Added memory recycling logic and other memory optimization measures, significantly reducing memory usage. The memory requirement for enabling all acceleration features except table acceleration (layout/formula/OCR) has been reduced from 16GB to 8GB, and the memory requirement for enabling all acceleration features has been reduced from 24GB to 10GB.
- Optimized configuration file feature switches, adding an independent formula detection switch to significantly improve speed and parsing results when formula detection is not needed.
- Integrated PDF-Extract-Kit 1.0:
- Added the self-developed
doclayout_yolomodel, which speeds up processing by more than 10 times compared to the original solution while maintaining similar parsing effects, and can be freely switched withlayoutlmv3via the configuration file. - Upgraded formula parsing to
unimernet 0.2.1, improving formula parsing accuracy while significantly reducing memory usage. - Due to the repository change for
PDF-Extract-Kit 1.0, you need to re-download the model. Please refer to How to Download Models for detailed steps.
- Added the self-developed
- 2024/09/27 Version 0.8.1 released, Fixed some bugs, and providing a localized deployment version of the online demo and the front-end interface.
- 2024/09/09: Version 0.8.0 released, supporting fast deployment with Dockerfile, and launching demos on Huggingface and Modelscope.
- 2024/08/30: Version 0.7.1 released, add paddle tablemaster table recognition option
- 2024/08/09: Version 0.7.0b1 released, simplified installation process, added table recognition functionality
- 2024/08/01: Version 0.6.2b1 released, optimized dependency conflict issues and installation documentation
- 2024/07/05: Initial open-source release
MinerU is a tool that converts PDFs into machine-readable formats (e.g., markdown, JSON), allowing for easy extraction into any format. MinerU was born during the pre-training process of InternLM. We focus on solving symbol conversion issues in scientific literature and hope to contribute to technological development in the era of large models. Compared to well-known commercial products, MinerU is still young. If you encounter any issues or if the results are not as expected, please submit an issue on issue and attach the relevant PDF.
pdf_zh_cn.mp4
- Remove headers, footers, footnotes, page numbers, etc., to ensure semantic coherence.
- Output text in human-readable order, suitable for single-column, multi-column, and complex layouts.
- Preserve the structure of the original document, including headings, paragraphs, lists, etc.
- Extract images, image descriptions, tables, table titles, and footnotes.
- Automatically recognize and convert formulas in the document to LaTeX format.
- Automatically recognize and convert tables in the document to HTML format.
- Automatically detect scanned PDFs and garbled PDFs and enable OCR functionality.
- OCR supports detection and recognition of 84 languages.
- Supports multiple output formats, such as multimodal and NLP Markdown, JSON sorted by reading order, and rich intermediate formats.
- Supports various visualization results, including layout visualization and span visualization, for efficient confirmation of output quality.
- Supports both CPU and GPU environments.
- Compatible with Windows, Linux, and Mac platforms.
If you encounter any installation issues, please first consult the FAQ.
If the parsing results are not as expected, refer to the Known Issues.
There are three different ways to experience MinerU:
Warning
Pre-installation Notice—Hardware and Software Environment Support
To ensure the stability and reliability of the project, we only optimize and test for specific hardware and software environments during development. This ensures that users deploying and running the project on recommended system configurations will get the best performance with the fewest compatibility issues.
By focusing resources on the mainline environment, our team can more efficiently resolve potential bugs and develop new features.
In non-mainline environments, due to the diversity of hardware and software configurations, as well as third-party dependency compatibility issues, we cannot guarantee 100% project availability. Therefore, for users who wish to use this project in non-recommended environments, we suggest carefully reading the documentation and FAQ first. Most issues already have corresponding solutions in the FAQ. We also encourage community feedback to help us gradually expand support.
| Operating System | |||||
| Ubuntu 22.04 LTS | Windows 10 / 11 | macOS 11+ | |||
| CPU | x86_64(unsupported ARM Linux) | x86_64(unsupported ARM Windows) | x86_64 / arm64 | ||
| Memory | 16GB or more, recommended 32GB+ | ||||
| Python Version | 3.10(Please make sure to create a Python 3.10 virtual environment using conda) | ||||
| Nvidia Driver Version | latest (Proprietary Driver) | latest | None | ||
| CUDA Environment | Automatic installation [12.1 (pytorch) + 11.8 (paddle)] | 11.8 (manual installation) + cuDNN v8.7.0 (manual installation) | None | ||
| GPU Hardware Support List | GPU VRAM 8GB or more | 2080~2080Ti / 3060Ti~3090Ti / 4060~4090 8G VRAM can enable all acceleration features |
None | ||
Stable Version (Stable version verified by QA):
Test Version (Synced with dev branch updates, testing new features):
conda create -n MinerU python=3.10
conda activate MinerU
pip install -U magic-pdf[full] --extra-index-url https://wheels.myhloli.comRefer to How to Download Model Files for detailed instructions.
After completing the 2. Download model weight files step, the script will automatically generate a magic-pdf.json file in the user directory and configure the default model path.
You can find the magic-pdf.json file in your 【user directory】.
Tip
The user directory for Windows is "C:\Users\username", for Linux it is "/home/username", and for macOS it is "/Users/username".
You can modify certain configurations in this file to enable or disable features, such as table recognition:
Note
If the following items are not present in the JSON, please manually add the required items and remove the comment content (standard JSON does not support comments).
{
// other config
"layout-config": {
"model": "layoutlmv3" // Please change to "doclayout_yolo" when using doclayout_yolo.
},
"formula-config": {
"mfd_model": "yolo_v8_mfd",
"mfr_model": "unimernet_small",
"enable": true // The formula recognition feature is enabled by default. If you need to disable it, please change the value here to "false".
},
"table-config": {
"model": "rapid_table", // Default to using "rapid_table", can be switched to "tablemaster" or "struct_eqtable".
"enable": false, // The table recognition feature is disabled by default. If you need to enable it, please change the value here to "true".
"max_time": 400
}
}If your device supports CUDA and meets the GPU requirements of the mainline environment, you can use GPU acceleration. Please select the appropriate guide based on your system:
- Ubuntu 22.04 LTS + GPU
- Windows 10/11 + GPU
- Quick Deployment with Docker
Important
Docker requires a GPU with at least 8GB of VRAM, and all acceleration features are enabled by default.
Before running this Docker, you can use the following command to check if your device supports CUDA acceleration on Docker.
docker run --rm --gpus=all nvidia/cuda:12.1.0-base-ubuntu22.04 nvidia-smiwget https://github.com/opendatalab/MinerU/raw/master/Dockerfile
docker build -t mineru:latest .
docker run --rm -it --gpus=all mineru:latest /bin/bash
magic-pdf --helpmagic-pdf --help
Usage: magic-pdf [OPTIONS]
Options:
-v, --version display the version and exit
-p, --path PATH local pdf filepath or directory [required]
-o, --output-dir PATH output local directory [required]
-m, --method [ocr|txt|auto] the method for parsing pdf. ocr: using ocr
technique to extract information from pdf. txt:
suitable for the text-based pdf only and
outperform ocr. auto: automatically choose the
best method for parsing pdf from ocr and txt.
without method specified, auto will be used by
default.
-l, --lang TEXT Input the languages in the pdf (if known) to
improve OCR accuracy. Optional. You should
input "Abbreviation" with language form url: ht
tps://paddlepaddle.github.io/PaddleOCR/latest/en
/ppocr/blog/multi_languages.html#5-support-languages-
and-abbreviations
-d, --debug BOOLEAN Enables detailed debugging information during
the execution of the CLI commands.
-s, --start INTEGER The starting page for PDF parsing, beginning
from 0.
-e, --end INTEGER The ending page for PDF parsing, beginning from
0.
--help Show this message and exit.
## show version
magic-pdf -v
## command line example
magic-pdf -p {some_pdf} -o {some_output_dir} -m auto{some_pdf} can be a single PDF file or a directory containing multiple PDFs.
The results will be saved in the {some_output_dir} directory. The output file list is as follows:
├── some_pdf.md # markdown file
├── images # directory for storing images
├── some_pdf_layout.pdf # layout diagram (Include layout reading order)
├── some_pdf_middle.json # MinerU intermediate processing result
├── some_pdf_model.json # model inference result
├── some_pdf_origin.pdf # original PDF file
├── some_pdf_spans.pdf # smallest granularity bbox position information diagram
└── some_pdf_content_list.json # Rich text JSON arranged in reading order
Tip
For more information about the output files, please refer to the Output File Description.
Processing files from local disk
image_writer = DiskReaderWriter(local_image_dir)
image_dir = str(os.path.basename(local_image_dir))
jso_useful_key = {"_pdf_type": "", "model_list": []}
pipe = UNIPipe(pdf_bytes, jso_useful_key, image_writer)
pipe.pipe_classify()
pipe.pipe_analyze()
pipe.pipe_parse()
md_content = pipe.pipe_mk_markdown(image_dir, drop_mode="none")Processing files from object storage
s3pdf_cli = S3ReaderWriter(pdf_ak, pdf_sk, pdf_endpoint)
image_dir = "s3://img_bucket/"
s3image_cli = S3ReaderWriter(img_ak, img_sk, img_endpoint, parent_path=image_dir)
pdf_bytes = s3pdf_cli.read(s3_pdf_path, mode=s3pdf_cli.MODE_BIN)
jso_useful_key = {"_pdf_type": "", "model_list": []}
pipe = UNIPipe(pdf_bytes, jso_useful_key, s3image_cli)
pipe.pipe_classify()
pipe.pipe_analyze()
pipe.pipe_parse()
md_content = pipe.pipe_mk_markdown(image_dir, drop_mode="none")For detailed implementation, refer to:
Derived projects include secondary development projects based on MinerU by project developers and community developers,
such as application interfaces based on Gradio, RAG based on llama, web demos similar to the official website, lightweight multi-GPU load balancing client/server ends, etc.
These projects may offer more features and a better user experience.
For specific deployment methods, please refer to the Derived Project README
TODO
- Reading order based on the model
- Recognition of
indexandlistin the main text - Table recognition
- Code block recognition in the main text
- Chemical formula recognition
- Geometric shape recognition
- Reading order is determined by the model based on the spatial distribution of readable content, and may be out of order in some areas under extremely complex layouts.
- Vertical text is not supported.
- Tables of contents and lists are recognized through rules, and some uncommon list formats may not be recognized.
- Only one level of headings is supported; hierarchical headings are not currently supported.
- Code blocks are not yet supported in the layout model.
- Comic books, art albums, primary school textbooks, and exercises cannot be parsed well.
- Table recognition may result in row/column recognition errors in complex tables.
- OCR recognition may produce inaccurate characters in PDFs of lesser-known languages (e.g., diacritical marks in Latin script, easily confused characters in Arabic script).
- Some formulas may not render correctly in Markdown.
This project currently uses PyMuPDF to achieve advanced functionality. However, since it adheres to the AGPL license, it may impose restrictions on certain usage scenarios. In future iterations, we plan to explore and replace it with a more permissive PDF processing library to enhance user-friendliness and flexibility.
- PDF-Extract-Kit
- DocLayout-YOLO
- StructEqTable
- RapidTable
- PaddleOCR
- PyMuPDF
- layoutreader
- fast-langdetect
- pdfminer.six
@misc{wang2024mineruopensourcesolutionprecise,
title={MinerU: An Open-Source Solution for Precise Document Content Extraction},
author={Bin Wang and Chao Xu and Xiaomeng Zhao and Linke Ouyang and Fan Wu and Zhiyuan Zhao and Rui Xu and Kaiwen Liu and Yuan Qu and Fukai Shang and Bo Zhang and Liqun Wei and Zhihao Sui and Wei Li and Botian Shi and Yu Qiao and Dahua Lin and Conghui He},
year={2024},
eprint={2409.18839},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2409.18839},
}
@article{he2024opendatalab,
title={Opendatalab: Empowering general artificial intelligence with open datasets},
author={He, Conghui and Li, Wei and Jin, Zhenjiang and Xu, Chao and Wang, Bin and Lin, Dahua},
journal={arXiv preprint arXiv:2407.13773},
year={2024}
}Magic-Doc Fast speed ppt/pptx/doc/docx/pdf extraction tool
Magic-HTML Mixed web page extraction tool