MinerU - An open-source multimodal non-OCR table recognition tool for document parsing and text extraction.

## Overview of MinerU MinerU is an open-source multimodal non-OCR table recognition tool primarily used for document parsing and text extraction. It supports 84 languages, multiple document layouts, and preserves the original structure of documents, including titles, paragraphs, and lists. It outputs in formats like Markdown and JSON and is compatible with both CPU and GPU environments. ## Overview of MinerU MinerU offers features such as multimodal non-OCR table recognition, document parsing, text extraction, removal of headers, footers, footnotes, and page numbers, semantic coherence, human-readable text output, support for single-column, multi-column, and complex layouts, structure preservation, extraction of images, image descriptions, tables, table captions, and footnotes, automatic formula conversion to LaTeX, 84 language support, multiple output formats, and compatibility with both CPU and GPU environments. ## Usage Methods of MinerU MinerU can be used via command line or Python API. The command-line usage involves the `magic-pdf` command with options for input path, output directory, method (OCR, text, auto), language, debug mode, and start and end pages. The Python API uses the `magic_pdf` library to process PDF, MS-Office, and image files, generating Markdown and JSON outputs. ## Language Support in MinerU MinerU supports the detection and recognition of 84 languages, making it versatile for global use. ## Output Formats of MinerU MinerU provides multiple output formats, including multimodal and NLP Markdown, and JSON sorted by reading order. ## Environment Compatibility of MinerU MinerU is compatible with both CPU and GPU environments, providing flexibility for deployment. ## Access Points for MinerU MinerU can be accessed through its project website on Hugging Face, its GitHub repository, and online demos available on mineru.net, Hugging Face Space, and ModelScope Studio. ## Primary Function of MinerU The primary function of MinerU is to extract content from documents, especially PDFs, and convert them into machine-readable formats like Markdown and JSON. ## Supported File Types in MinerU In addition to PDFs, MinerU supports MS-Office documents (ppt, pptx, doc, docx) and images (png, jpg). ## Significance of MinerU in Document Analysis MinerU addresses common challenges in document analysis, such as symbol conversion in scientific literature, making it a valuable tool in the large model era. Its ability to handle diverse document types and preserve structure enhances its utility in academic and technical fields. ### Citation sources: - [MinerU](https://github.com/opendatalab/MinerU) - Official URL Updated: 2025-03-28