README_zh-CN.md 36 KB
Newer Older
xuchao's avatar
xuchao committed
1
2
<div align="center" xmlns="http://www.w3.org/1999/html">
<!-- logo -->
徐超's avatar
徐超 committed
3
<p align="center">
4
  <img src="docs/images/MinerU-logo.png" width="300px" style="vertical-align:middle;">
徐超's avatar
徐超 committed
5
</p>
赵小蒙's avatar
赵小蒙 committed
6

xuchao's avatar
xuchao committed
7
<!-- icon -->
8

赵小蒙's avatar
赵小蒙 committed
9
10
11
[![stars](https://img.shields.io/github/stars/opendatalab/MinerU.svg)](https://github.com/opendatalab/MinerU)
[![forks](https://img.shields.io/github/forks/opendatalab/MinerU.svg)](https://github.com/opendatalab/MinerU)
[![open issues](https://img.shields.io/github/issues-raw/opendatalab/MinerU)](https://github.com/opendatalab/MinerU/issues)
myhloli's avatar
myhloli committed
12
13
14
15
[![issue resolution](https://img.shields.io/github/issues-closed-raw/opendatalab/MinerU)](https://github.com/opendatalab/MinerU/issues)
[![PyPI version](https://badge.fury.io/py/magic-pdf.svg)](https://badge.fury.io/py/magic-pdf)
[![Downloads](https://static.pepy.tech/badge/magic-pdf)](https://pepy.tech/project/magic-pdf)
[![Downloads](https://static.pepy.tech/badge/magic-pdf/month)](https://pepy.tech/project/magic-pdf)
16

17
[![OpenDataLab](https://img.shields.io/badge/Demo_on_OpenDataLab-blue?logo=data:image/svg+xml;base64,PHN2ZyB3aWR0aD0iMTM0IiBoZWlnaHQ9IjEzNCIgeG1sbnM9Imh0dHA6Ly93d3cudzMub3JnLzIwMDAvc3ZnIj48cGF0aCBkPSJtMTIyLDljMCw1LTQsOS05LDlzLTktNC05LTksNC05LDktOSw5LDQsOSw5eiIgZmlsbD0idXJsKCNhKSIvPjxwYXRoIGQ9Im0xMjIsOWMwLDUtNCw5LTksOXMtOS00LTktOSw0LTksOS05LDksNCw5LDl6IiBmaWxsPSIjMDEwMTAxIi8+PHBhdGggZD0ibTkxLDE4YzAsNS00LDktOSw5cy05LTQtOS05LDQtOSw5LTksOSw0LDksOXoiIGZpbGw9InVybCgjYikiLz48cGF0aCBkPSJtOTEsMThjMCw1LTQsOS05LDlzLTktNC05LTksNC05LDktOSw5LDQsOSw5eiIgZmlsbD0iIzAxMDEwMSIvPjxwYXRoIGZpbGwtcnVsZT0iZXZlbm9kZCIgY2xpcC1ydWxlPSJldmVub2RkIiBkPSJtMzksNjJjMCwxNiw4LDMwLDIwLDM4LDctNiwxMi0xNiwxMi0yNlY0OWMwLTQsMy03LDYtOGw0Ni0xMmM1LTEsMTEsMywxMSw4djMxYzAsMzctMzAsNjYtNjYsNjYtMzcsMC02Ni0zMC02Ni02NlY0NmMwLTQsMy03LDYtOGwyMC02YzUtMSwxMSwzLDExLDh2MjF6bS0yOSw2YzAsMTYsNiwzMCwxNyw0MCwzLDEsNSwxLDgsMSw1LDAsMTAtMSwxNS0zQzM3LDk1LDI5LDc5LDI5LDYyVjQybC0xOSw1djIweiIgZmlsbD0idXJsKCNjKSIvPjxwYXRoIGZpbGwtcnVsZT0iZXZlbm9kZCIgY2xpcC1ydWxlPSJldmVub2RkIiBkPSJtMzksNjJjMCwxNiw4LDMwLDIwLDM4LDctNiwxMi0xNiwxMi0yNlY0OWMwLTQsMy03LDYtOGw0Ni0xMmM1LTEsMTEsMywxMSw4djMxYzAsMzctMzAsNjYtNjYsNjYtMzcsMC02Ni0zMC02Ni02NlY0NmMwLTQsMy03LDYtOGwyMC02YzUtMSwxMSwzLDExLDh2MjF6bS0yOSw2YzAsMTYsNiwzMCwxNyw0MCwzLDEsNSwxLDgsMSw1LDAsMTAtMSwxNS0zQzM3LDk1LDI5LDc5LDI5LDYyVjQybC0xOSw1djIweiIgZmlsbD0iIzAxMDEwMSIvPjxkZWZzPjxsaW5lYXJHcmFkaWVudCBpZD0iYSIgeDE9Ijg0IiB5MT0iNDEiIHgyPSI3NSIgeTI9IjEyMCIgZ3JhZGllbnRVbml0cz0idXNlclNwYWNlT25Vc2UiPjxzdG9wIHN0b3AtY29sb3I9IiNmZmYiLz48c3RvcCBvZmZzZXQ9IjEiIHN0b3AtY29sb3I9IiMyZTJlMmUiLz48L2xpbmVhckdyYWRpZW50PjxsaW5lYXJHcmFkaWVudCBpZD0iYiIgeDE9Ijg0IiB5MT0iNDEiIHgyPSI3NSIgeTI9IjEyMCIgZ3JhZGllbnRVbml0cz0idXNlclNwYWNlT25Vc2UiPjxzdG9wIHN0b3AtY29sb3I9IiNmZmYiLz48c3RvcCBvZmZzZXQ9IjEiIHN0b3AtY29sb3I9IiMyZTJlMmUiLz48L2xpbmVhckdyYWRpZW50PjxsaW5lYXJHcmFkaWVudCBpZD0iYyIgeDE9Ijg0IiB5MT0iNDEiIHgyPSI3NSIgeTI9IjEyMCIgZ3JhZGllbnRVbml0cz0idXNlclNwYWNlT25Vc2UiPjxzdG9wIHN0b3AtY29sb3I9IiNmZmYiLz48c3RvcCBvZmZzZXQ9IjEiIHN0b3AtY29sb3I9IiMyZTJlMmUiLz48L2xpbmVhckdyYWRpZW50PjwvZGVmcz48L3N2Zz4=&labelColor=white)](https://mineru.net/OpenSourceTools/Extractor?source=github)
myhloli's avatar
myhloli committed
18
[![ModelScope](https://img.shields.io/badge/Demo_on_ModelScope-purple?logo=data:image/svg+xml;base64,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&labelColor=white)](https://www.modelscope.cn/studios/OpenDataLab/MinerU)
myhloli's avatar
myhloli committed
19
[![HuggingFace](https://img.shields.io/badge/Demo_on_HuggingFace-yellow.svg?logo=data:image/png;base64,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&labelColor=white)](https://huggingface.co/spaces/opendatalab/MinerU)
myhloli's avatar
myhloli committed
20
[![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/gist/myhloli/3b3a00a4a0a61577b6c30f989092d20d/mineru_demo.ipynb)
21
[![Paper](https://img.shields.io/badge/Paper-arXiv-green)](https://arxiv.org/abs/2409.18839)
22

myhloli's avatar
myhloli committed
23

myhloli's avatar
myhloli committed
24
<a href="https://trendshift.io/repositories/11174" target="_blank"><img src="https://trendshift.io/api/badge/repositories/11174" alt="opendatalab%2FMinerU | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
赵小蒙's avatar
赵小蒙 committed
25

xuchao's avatar
xuchao committed
26
<!-- language -->
赵小蒙's avatar
赵小蒙 committed
27

28
[English](README.md) | [简体中文](README_zh-CN.md)
赵小蒙's avatar
赵小蒙 committed
29

xuchao's avatar
xuchao committed
30
<!-- hot link -->
31

徐超's avatar
徐超 committed
32
33
<p align="center">
<a href="https://github.com/opendatalab/PDF-Extract-Kit">PDF-Extract-Kit: 高质量PDF解析工具箱</a>🔥🔥🔥
34
35
<br>
<br>
36
<a href="https://mineru.net/client?source=github">更便捷的使用方式:MinerU桌面端。无需编程,无需登录,图形界面,简单交互,畅用无忧。</a>🚀🚀🚀
37

徐超's avatar
徐超 committed
38
39
</p>

xuchao's avatar
xuchao committed
40
<!-- join us -->
41

徐超's avatar
徐超 committed
42
<p align="center">
43
    👋 join us on <a href="https://discord.gg/Tdedn9GTXq" target="_blank">Discord</a> and <a href="http://mineru.space/s/V85Yl" target="_blank">WeChat</a>
徐超's avatar
徐超 committed
44
</p>
赵小蒙's avatar
赵小蒙 committed
45

xuchao's avatar
xuchao committed
46
</div>
赵小蒙's avatar
赵小蒙 committed
47

xuchao's avatar
xuchao committed
48
# 更新记录
49
50
- 2025/04/03 1.3.0 发布,在这个版本我们做出了许多改变:
  - 安装与兼容性优化
51
    - 通过使用paddleocr2torch,完全替代了paddle框架以及paddleocr在项目中的使用,解决了paddle和torch的冲突问题
52
53
54
55
    - 通过移除layout中layoutlmv3的使用,解决了由`detectron2`导致的兼容问题
    - torch版本兼容扩展到2.2~2.6
    - cuda兼容扩展到11.8~12.6(cuda版本由torch决定),解决部分用户50系显卡与H系显卡的兼容问题
    - python兼容版本扩展到3.10~3.12,解决了在非3.10环境下安装时自动降级到0.6.1的问题
56
    - 优化离线部署流程,部署成功后不需要联网下载任何模型文件
57
  - 性能优化(与1.0.1版本相比,公式解析速度最高提升超过1400%,整体解析速度提升超过500%)
58
59
    - 通过支持多个pdf文件的batch处理([脚本样例](demo/batch_demo.py)),提升了批量小文件的解析速度
    - 通过优化mfr模型的加载和使用,降低了显存占用并提升了解析速度(需重新执行[模型下载流程](docs/how_to_download_models_zh_cn.md)以获得模型文件的增量更新)
60
61
62
63
    - 优化显存占用,最低仅需6GB即可运行本项目
    - 优化了在mps设备上的运行速度
  - 解析效果优化
    - mfr模型更新到unimernet(2503),解决多行公式中换行丢失的问题
64
65
66
67
- 2025/03/03 1.2.1 发布,修复了一些问题:
  - 修复在字母与数字的全角转半角操作时对标点符号的影响
  - 修复在某些情况下caption的匹配不准确问题
  - 修复在某些情况下的公式span丢失问题
68
69
70
71
72
73
74
75
76
77
- 2025/02/24 1.2.0 发布,这个版本我们修复了一些问题,提升了解析的效率与精度:
  - 性能优化 
    - auto模式下pdf文档的分类速度提升
    - 在华为昇腾 NPU 加速模式下,添加高性能插件支持,常见场景下端到端加速可达 300% [申请链接](https://aicarrier.feishu.cn/share/base/form/shrcnb10VaoNQB8kQPA8DEfZC6d)
  - 解析优化
    - 优化对包含水印文档的解析逻辑,显著提升包含水印文档的解析效果
    - 改进了单页内多个图像/表格与caption的匹配逻辑,提升了复杂布局下图文匹配的准确性
  - 问题修复
    - 修复在某些情况下图片/表格span被填充进textblock导致的异常
    - 修复在某些情况下标题block为空的问题
myhloli's avatar
myhloli committed
78
- 2025/01/22 1.1.0 发布,在这个版本我们重点提升了解析的精度与效率:
79
80
81
82
83
84
  - 模型能力升级(需重新执行[模型下载流程](docs/how_to_download_models_zh_cn.md)以获得模型文件的增量更新) 
    - 布局识别模型升级到最新的`doclayout_yolo(2501)`模型,提升了layout识别精度
    - 公式解析模型升级到最新的`unimernet(2501)`模型,提升了公式识别精度
  - 性能优化
    - 在配置满足一定条件(显存16GB+)的设备上,通过优化资源占用和重构处理流水线,整体解析速度提升50%以上
  - 解析效果优化
85
    - 在线demo([mineru.net](https://mineru.net/OpenSourceTools/Extractor)/[huggingface](https://huggingface.co/spaces/opendatalab/MinerU)/[modelscope](https://www.modelscope.cn/studios/OpenDataLab/MinerU))上新增标题分级功能(测试版本,默认开启),支持对标题进行分级,提升文档结构化程度
86
- 2025/01/10 1.0.1 发布,这是我们的第一个正式版本,在这个版本中,我们通过大量重构带来了全新的API接口和更广泛的兼容性,以及全新的自动语言识别功能:
87
88
89
90
  - 全新API接口 
    - 对于数据侧API,我们引入了Dataset类,旨在提供一个强大而灵活的数据处理框架。该框架当前支持包括图像(.jpg及.png)、PDF、Word(.doc及.docx)、以及PowerPoint(.ppt及.pptx)在内的多种文档格式,确保了从简单到复杂的数据处理任务都能得到有效的支持。
    - 针对用户侧API,我们将MinerU的处理流程精心设计为一系列可组合的Stage阶段。每个Stage代表了一个特定的处理步骤,用户可以根据自身需求自由地定义新的Stage,并通过创造性地组合这些阶段来定制专属的数据处理流程。
  - 更广泛的兼容性适配
91
    - 通过优化依赖环境和配置项,确保在ARM架构的Linux系统上能够稳定高效运行。
92
93
94
    - 深度适配华为昇腾NPU加速,积极响应信创要求,提供自主可控的高性能计算能力,助力人工智能应用平台的国产化应用与发展。[NPU加速教程](docs/README_Ascend_NPU_Acceleration_zh_CN.md)
  - 自动语言识别
    - 通过引入全新的语言识别模型, 在文档解析中将`lang`配置为`auto`,即可自动选择合适的OCR语言模型,提升扫描类文档解析的准确性。
myhloli's avatar
myhloli committed
95
96
97
- 2024/11/22 0.10.0发布,通过引入混合OCR文本提取能力,
  - 在公式密集、span区域不规范、部分文本使用图像表现等复杂文本分布场景下获得解析效果的显著提升
  - 同时具备文本模式内容提取准确、速度更快与OCR模式span/line区域识别更准的双重优势
myhloli's avatar
myhloli committed
98
- 2024/11/15 0.9.3发布,为表格识别功能接入了[RapidTable](https://github.com/RapidAI/RapidTable),单表解析速度提升10倍以上,准确率更高,显存占用更低
99
- 2024/11/06 0.9.2发布,为表格识别功能接入了[StructTable-InternVL2-1B](https://huggingface.co/U4R/StructTable-InternVL2-1B)模型
100
101
102
103
- 2024/10/31 0.9.0发布,这是我们进行了大量代码重构的全新版本,解决了众多问题,提升了性能,降低了硬件需求,并提供了更丰富的易用性:
  - 重构排序模块代码,使用 [layoutreader](https://github.com/ppaanngggg/layoutreader) 进行阅读顺序排序,确保在各种排版下都能实现极高准确率
  - 重构段落拼接模块,在跨栏、跨页、跨图、跨表情况下均能实现良好的段落拼接效果
  - 重构列表和目录识别功能,极大提升列表块和目录块识别的准确率及对应文本段落的解析效果
104
  - 重构图、表与描述性文本的匹配逻辑,大幅提升 caption 和 footnote 与图表的匹配准确率,并将描述性文本的丢失率降至接近0
105
106
107
108
109
110
  - 增加 OCR 的多语言支持,支持 84 种语言的检测与识别,语言支持列表详见 [OCR 语言支持列表](https://paddlepaddle.github.io/PaddleOCR/latest/ppocr/blog/multi_languages.html#5)
  - 增加显存回收逻辑及其他显存优化措施,大幅降低显存使用需求。开启除表格加速外的全部加速功能(layout/公式/OCR)的显存需求从16GB降至8GB,开启全部加速功能的显存需求从24GB降至10GB
  - 优化配置文件的功能开关,增加独立的公式检测开关,无需公式检测时可大幅提升速度和解析效果
  - 集成 [PDF-Extract-Kit 1.0](https://github.com/opendatalab/PDF-Extract-Kit)
    - 加入自研的 `doclayout_yolo` 模型,在相近解析效果情况下比原方案提速10倍以上,可通过配置文件与 `layoutlmv3` 自由切换
    - 公式解析升级至 `unimernet 0.2.1`,在提升公式解析准确率的同时,大幅降低显存需求
111
    -`PDF-Extract-Kit 1.0` 更换仓库,需要重新下载模型,步骤详见 [如何下载模型](docs/how_to_download_models_zh_cn.md)
sfk's avatar
sfk committed
112
- 2024/09/27 0.8.1发布,修复了一些bug,同时提供了[在线demo](https://opendatalab.com/OpenSourceTools/Extractor/PDF/)[本地化部署版本](projects/web_demo/README_zh-CN.md)[前端界面](projects/web/README_zh-CN.md)
drunkpig's avatar
drunkpig committed
113
- 2024/09/09 0.8.0发布,支持Dockerfile快速部署,同时上线了huggingface、modelscope demo
114
- 2024/08/30 0.7.1发布,集成了paddle tablemaster表格识别功能
xuchao's avatar
xuchao committed
115
116
117
118
119
- 2024/08/09 0.7.0b1发布,简化安装步骤提升易用性,加入表格识别功能
- 2024/08/01 0.6.2b1发布,优化了依赖冲突问题和安装文档
- 2024/07/05 首次开源

<!-- TABLE OF CONTENT -->
120

xuchao's avatar
xuchao committed
121
122
123
124
125
126
127
128
129
130
131
132
<details open="open">
  <summary><h2 style="display: inline-block">文档目录</h2></summary>
  <ol>
    <li>
      <a href="#mineru">MinerU</a>
      <ul>
        <li><a href="#项目简介">项目简介</a></li>
        <li><a href="#主要功能">主要功能</a></li>
        <li><a href="#快速开始">快速开始</a>
            <ul>
            <li><a href="#在线体验">在线体验</a></li>
            <li><a href="#使用CPU快速体验">使用CPU快速体验</a></li>
myhloli's avatar
myhloli committed
133
            <li><a href="#使用GPU">使用GPU</a></li>
134
            <li><a href="#使用NPU">使用NPU</a></li>
xuchao's avatar
xuchao committed
135
136
137
138
            </ul>
        </li>
        <li><a href="#使用">使用方式</a>
            <ul>
myhloli's avatar
myhloli committed
139
            <li><a href="#命令行">命令行</a></li>
xuchao's avatar
xuchao committed
140
            <li><a href="#api">API</a></li>
141
            <li><a href="#部署衍生项目">部署衍生项目</a></li>
xuchao's avatar
xuchao committed
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
            <li><a href="#二次开发">二次开发</a></li>
            </ul>
        </li>
      </ul>
    </li>
    <li><a href="#todo">TODO</a></li>
    <li><a href="#known-issues">Known Issues</a></li>
    <li><a href="#faq">FAQ</a></li>
    <li><a href="#all-thanks-to-our-contributors">Contributors</a></li>
    <li><a href="#license-information">License Information</a></li>
    <li><a href="#acknowledgments">Acknowledgements</a></li>
    <li><a href="#citation">Citation</a></li>
    <li><a href="#star-history">Star History</a></li>
    <li><a href="#magic-doc">magic-doc快速提取PPT/DOC/PDF</a></li>
    <li><a href="#magic-html">magic-html提取混合网页内容</a></li>
    <li><a href="#links">Links</a></li>
  </ol>
</details>

# MinerU
162

xuchao's avatar
xuchao committed
163
## 项目简介
164

xuchao's avatar
xuchao committed
165
166
167
MinerU是一款将PDF转化为机器可读格式的工具(如markdown、json),可以很方便地抽取为任意格式。
MinerU诞生于[书生-浦语](https://github.com/InternLM/InternLM)的预训练过程中,我们将会集中精力解决科技文献中的符号转化问题,希望在大模型时代为科技发展做出贡献。
相比国内外知名商用产品MinerU还很年轻,如果遇到问题或者结果不及预期请到[issue](https://github.com/opendatalab/MinerU/issues)提交问题,同时**附上相关PDF**
myhloli's avatar
myhloli committed
168

Xiaomeng Zhao's avatar
Xiaomeng Zhao committed
169
https://github.com/user-attachments/assets/4bea02c9-6d54-4cd6-97ed-dff14340982c
myhloli's avatar
myhloli committed
170

myhloli's avatar
myhloli committed
171
172
173
174
175
176
177
## 主要功能

- 删除页眉、页脚、脚注、页码等元素,确保语义连贯
- 输出符合人类阅读顺序的文本,适用于单栏、多栏及复杂排版
- 保留原文档的结构,包括标题、段落、列表等
- 提取图像、图片描述、表格、表格标题及脚注
- 自动识别并转换文档中的公式为LaTeX格式
178
- 自动识别并转换文档中的表格为HTML格式
myhloli's avatar
myhloli committed
179
180
181
182
- 自动检测扫描版PDF和乱码PDF,并启用OCR功能
- OCR支持84种语言的检测与识别
- 支持多种输出格式,如多模态与NLP的Markdown、按阅读顺序排序的JSON、含有丰富信息的中间格式等
- 支持多种可视化结果,包括layout可视化、span可视化等,便于高效确认输出效果与质检
183
- 支持纯CPU环境运行,并支持 GPU(CUDA)/NPU(CANN)/MPS 加速
myhloli's avatar
myhloli committed
184
- 兼容Windows、Linux和Mac平台
赵小蒙's avatar
update:  
赵小蒙 committed
185

xuchao's avatar
xuchao committed
186
187
## 快速开始

myhloli's avatar
myhloli committed
188
189
190
如果遇到任何安装问题,请先查询 <a href="#faq">FAQ</a> </br>
如果遇到解析效果不及预期,参考 <a href="#known-issues">Known Issues</a></br>
有3种不同方式可以体验MinerU的效果:
191

xuchao's avatar
xuchao committed
192
193
- [在线体验(无需任何安装)](#在线体验)
- [使用CPU快速体验(Windows,Linux,Mac)](#使用cpu快速体验)
194
- 使用 CUDA/CANN/MPS 加速推理 
195
  - [Linux/Windows + CUDA](#使用gpu)
196
197
  - [Linux + CANN](#使用npu)
  - [MacOS + MPS](#使用mps)
myhloli's avatar
myhloli committed
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213


> [!WARNING]
> **安装前必看——软硬件环境支持说明**
> 
> 为了确保项目的稳定性和可靠性,我们在开发过程中仅对特定的软硬件环境进行优化和测试。这样当用户在推荐的系统配置上部署和运行项目时,能够获得最佳的性能表现和最少的兼容性问题。
>
> 通过集中资源和精力于主线环境,我们团队能够更高效地解决潜在的BUG,及时开发新功能。
>
> 在非主线环境中,由于硬件、软件配置的多样性,以及第三方依赖项的兼容性问题,我们无法100%保证项目的完全可用性。因此,对于希望在非推荐环境中使用本项目的用户,我们建议先仔细阅读文档以及FAQ,大多数问题已经在FAQ中有对应的解决方案,除此之外我们鼓励社区反馈问题,以便我们能够逐步扩大支持范围。

<table>
    <tr>
        <td colspan="3" rowspan="2">操作系统</td>
    </tr>
    <tr>
214
        <td>Linux after 2019</td>
myhloli's avatar
myhloli committed
215
216
217
218
219
        <td>Windows 10 / 11</td>
        <td>macOS 11+</td>
    </tr>
    <tr>
        <td colspan="3">CPU</td>
220
        <td>x86_64 / arm64</td>
myhloli's avatar
myhloli committed
221
222
223
224
225
226
227
        <td>x86_64(暂不支持ARM Windows)</td>
        <td>x86_64 / arm64</td>
    </tr>
    <tr>
        <td colspan="3">内存</td>
        <td colspan="3">大于等于16GB,推荐32G以上</td>
    </tr>
228
229
230
231
    <tr>
        <td colspan="3">存储空间</td>
        <td colspan="3">大于等于20GB,推荐使用SSD以获得最佳性能</td>
    </tr>
myhloli's avatar
myhloli committed
232
233
    <tr>
        <td colspan="3">python版本</td>
234
        <td colspan="3">>=3.9,<=3.12</td>
myhloli's avatar
myhloli committed
235
236
237
238
239
240
241
242
243
    </tr>
    <tr>
        <td colspan="3">Nvidia Driver 版本</td>
        <td>latest(专有驱动)</td>
        <td>latest</td>
        <td>None</td>
    </tr>
    <tr>
        <td colspan="3">CUDA环境</td>
244
245
        <td>11.8/12.4/12.6</td>
        <td>11.8/12.4/12.6</td>
myhloli's avatar
myhloli committed
246
247
        <td>None</td>
    </tr>
248
249
250
251
252
253
    <tr>
        <td colspan="3">CANN环境(NPU支持)</td>
        <td>8.0+(Ascend 910b)</td>
        <td>None</td>
        <td>None</td>
    </tr>
myhloli's avatar
myhloli committed
254
    <tr>
255
256
        <td rowspan="2">GPU/MPS 硬件支持列表</td>
        <td colspan="2">显存6G以上</td>
257
        <td colspan="2">
258
259
260
        Volta(2017)及之后生产的全部带Tensor Core的GPU <br>
        6G显存及以上</td>
        <td rowspan="2">apple slicon</td>
myhloli's avatar
myhloli committed
261
262
    </tr>
</table>
xuchao's avatar
xuchao committed
263
264
265

### 在线体验

Xiaomeng Zhao's avatar
Xiaomeng Zhao committed
266
同步dev分支更新:
267

Xiaomeng Zhao's avatar
Xiaomeng Zhao committed
268
[![OpenDataLab](https://img.shields.io/badge/Demo_on_OpenDataLab-blue?logo=data:image/svg+xml;base64,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&labelColor=white)](https://mineru.net/OpenSourceTools/Extractor?source=github)
269
[![ModelScope](https://img.shields.io/badge/Demo_on_ModelScope-purple?logo=data:image/svg+xml;base64,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&labelColor=white)](https://www.modelscope.cn/studios/OpenDataLab/MinerU)
Xiaomeng Zhao's avatar
Xiaomeng Zhao committed
270
[![HuggingFace](https://img.shields.io/badge/Demo_on_HuggingFace-yellow.svg?logo=data:image/png;base64,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&labelColor=white)](https://huggingface.co/spaces/opendatalab/MinerU)
xuchao's avatar
xuchao committed
271
272
273
274

### 使用CPU快速体验

#### 1. 安装magic-pdf
275

276
277
> [!NOTE]
> 最新版本国内镜像源同步可能会有延迟,请耐心等待
278

279
```bash
280
conda create -n mineru 'python<3.13' -y
281
conda activate mineru
282
pip install -U "magic-pdf[full]" -i https://mirrors.aliyun.com/pypi/simple
283
```
284

285
#### 2. 下载模型权重文件
赵小蒙's avatar
赵小蒙 committed
286

xuchao's avatar
xuchao committed
287
详细参考 [如何下载模型文件](docs/how_to_download_models_zh_cn.md)
288

289
#### 3. 修改配置文件以进行额外配置
290

291
292
完成[2. 下载模型权重文件](#2-下载模型权重文件)步骤后,脚本会自动生成用户目录下的magic-pdf.json文件,并自动配置默认模型路径。
您可在【用户目录】下找到magic-pdf.json文件。
293
294

> [!TIP]
295
296
> windows的用户目录为 "C:\\Users\\用户名", linux用户目录为 "/home/用户名", macOS用户目录为 "/Users/用户名"

297
您可修改该文件中的部分配置实现功能的开关,如表格识别功能:
298

299
> [!NOTE]
300
>如json内没有如下项目,请手动添加需要的项目,并删除注释内容(标准json不支持注释)
301

302
303
```json
{
304
305
    // other config
    "layout-config": {
306
        "model": "doclayout_yolo" 
307
308
309
310
311
312
313
    },
    "formula-config": {
        "mfd_model": "yolo_v8_mfd",
        "mfr_model": "unimernet_small",
        "enable": true  // 公式识别功能默认是开启的,如果需要关闭请修改此处的值为"false"
    },
    "table-config": {
314
315
        "model": "rapid_table",
        "sub_model": "slanet_plus",
316
        "enable": true, // 表格识别功能默认是开启的,如果需要关闭请修改此处的值为"false"
xuchao's avatar
xuchao committed
317
318
        "max_time": 400
    }
319
320
321
}
```

myhloli's avatar
myhloli committed
322
323
324
325
326
327
328
329
### 使用GPU

如果您的设备支持CUDA,且满足主线环境中的显卡要求,则可以使用GPU加速,请根据自己的系统选择适合的教程:

- [Ubuntu22.04LTS + GPU](docs/README_Ubuntu_CUDA_Acceleration_zh_CN.md)
- [Windows10/11 + GPU](docs/README_Windows_CUDA_Acceleration_zh_CN.md)
- 使用Docker快速部署
> [!IMPORTANT]
330
> Docker 需设备gpu显存大于等于6GB,默认开启所有加速功能
myhloli's avatar
myhloli committed
331
332
333
334
335
336
337
> 
> 运行本docker前可以通过以下命令检测自己的设备是否支持在docker上使用CUDA加速
> 
> ```bash
> docker run --rm --gpus=all nvidia/cuda:12.1.0-base-ubuntu22.04 nvidia-smi
> ```
  ```bash
338
  wget https://gcore.jsdelivr.net/gh/opendatalab/MinerU@master/docker/china/Dockerfile -O Dockerfile
myhloli's avatar
myhloli committed
339
  docker build -t mineru:latest .
340
  docker run -it --name mineru --gpus=all mineru:latest /bin/bash -c "echo 'source /opt/mineru_venv/bin/activate' >> ~/.bashrc && exec bash"
myhloli's avatar
myhloli committed
341
342
  magic-pdf --help
  ```
343
344
345
### 使用NPU

如果您的设备存在NPU加速硬件,则可以通过以下教程使用NPU加速:
346

347
[NPU加速教程](docs/README_Ascend_NPU_Acceleration_zh_CN.md)
348

349
### 使用MPS
350
如果您的设备使用Apple silicon 芯片,您可以开启mps加速:
351
352
353
354
355
356
357
358
359
360
361
362

您可以通过在 `magic-pdf.json` 配置文件中将 `device-mode` 参数设置为 `mps` 来启用 MPS 加速。

```json
{
    // other config
    "device-mode": "mps"
}
```



xuchao's avatar
xuchao committed
363
## 使用
364

myhloli's avatar
myhloli committed
365
366
### 命令行

367
[通过命令行使用MinerU](https://mineru.readthedocs.io/en/latest/user_guide/usage/command_line.html)
myhloli's avatar
myhloli committed
368
369
370
371

> [!TIP]
> 更多有关输出文件的信息,请参考[输出文件说明](docs/output_file_zh_cn.md)

xuchao's avatar
xuchao committed
372
373
### API

374
[通过Python代码调用MinerU](https://mineru.readthedocs.io/en/latest/user_guide/usage/api.html)
赵小蒙's avatar
赵小蒙 committed
375

myhloli's avatar
myhloli committed
376

377
378
379
380
381
382
383
384
### 部署衍生项目

衍生项目包含项目开发者和社群开发者们基于MinerU的二次开发项目,
例如基于Gradio的应用界面、基于llama的RAG、官网同款web demo、轻量级的多卡负载均衡c/s端等,
这些项目可能会提供更多的功能和更好的用户体验。
具体部署方式请参考 [衍生项目readme](projects/README_zh-CN.md)


xuchao's avatar
xuchao committed
385
### 二次开发
386

xuchao's avatar
xuchao committed
387
TODO
388

xuchao's avatar
xuchao committed
389
# TODO
赵小蒙's avatar
赵小蒙 committed
390

391
392
393
- [x] 基于模型的阅读顺序  
- [x] 正文中目录、列表识别  
- [x] 表格识别
myhloli's avatar
myhloli committed
394
- [x] 标题分级
395
396
397
- [ ] 正文中代码块识别
- [ ] [化学式识别](docs/chemical_knowledge_introduction/introduction.pdf)
- [ ] 几何图形识别
赵小蒙's avatar
赵小蒙 committed
398

myhloli's avatar
myhloli committed
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
# Known Issues

- 阅读顺序基于模型对可阅读内容在空间中的分布进行排序,在极端复杂的排版下可能会部分区域乱序
- 不支持竖排文字
- 目录和列表通过规则进行识别,少部分不常见的列表形式可能无法识别
- 代码块在layout模型里还没有支持
- 漫画书、艺术图册、小学教材、习题尚不能很好解析
- 表格识别在复杂表格上可能会出现行/列识别错误
- 在小语种PDF上,OCR识别可能会出现字符不准确的情况(如拉丁文的重音符号、阿拉伯文易混淆字符等)
- 部分公式可能会无法在markdown中渲染

# FAQ

[常见问题](docs/FAQ_zh_cn.md)


[FAQ](docs/FAQ_en_us.md)
myhloli's avatar
myhloli committed
416

xuchao's avatar
xuchao committed
417
# All Thanks To Our Contributors
赵小蒙's avatar
赵小蒙 committed
418

419
<a href="https://github.com/opendatalab/MinerU/graphs/contributors">
赵小蒙's avatar
赵小蒙 committed
420
421
422
  <img src="https://contrib.rocks/image?repo=opendatalab/MinerU" />
</a>

xuchao's avatar
xuchao committed
423
# License Information
赵小蒙's avatar
赵小蒙 committed
424
425
426
427
428

[LICENSE.md](LICENSE.md)

本项目目前采用PyMuPDF以实现高级功能,但因其遵循AGPL协议,可能对某些使用场景构成限制。未来版本迭代中,我们计划探索并替换为许可条款更为宽松的PDF处理库,以提升用户友好度及灵活性。

xuchao's avatar
xuchao committed
429
# Acknowledgments
430

xuchao's avatar
xuchao committed
431
- [PDF-Extract-Kit](https://github.com/opendatalab/PDF-Extract-Kit)
432
- [DocLayout-YOLO](https://github.com/opendatalab/DocLayout-YOLO)
xuchao's avatar
xuchao committed
433
- [StructEqTable](https://github.com/UniModal4Reasoning/StructEqTable-Deploy)
434
- [RapidTable](https://github.com/RapidAI/RapidTable)
赵小蒙's avatar
赵小蒙 committed
435
- [PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR)
436
437
- [RapidOCR](https://github.com/RapidAI/RapidOCR)
- [PaddleOCR2Pytorch](https://github.com/frotms/PaddleOCR2Pytorch)
赵小蒙's avatar
赵小蒙 committed
438
- [PyMuPDF](https://github.com/pymupdf/PyMuPDF)
439
- [layoutreader](https://github.com/ppaanngggg/layoutreader)
赵小蒙's avatar
赵小蒙 committed
440
441
- [fast-langdetect](https://github.com/LlmKira/fast-langdetect)
- [pdfminer.six](https://github.com/pdfminer/pdfminer.six)
赵小蒙's avatar
赵小蒙 committed
442

xuchao's avatar
xuchao committed
443
# Citation
赵小蒙's avatar
赵小蒙 committed
444
445

```bibtex
446
447
448
449
450
451
452
453
454
455
@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}, 
}

xuchao's avatar
xuchao committed
456
457
458
459
460
461
@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}
}
赵小蒙's avatar
赵小蒙 committed
462
463
464
```

# Star History
赵小蒙's avatar
赵小蒙 committed
465

赵小蒙's avatar
赵小蒙 committed
466
467
468
469
470
471
<a>
 <picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=opendatalab/MinerU&type=Date&theme=dark" />
   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=opendatalab/MinerU&type=Date" />
   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=opendatalab/MinerU&type=Date" />
 </picture>
Xiaomeng Zhao's avatar
Xiaomeng Zhao committed
472
</a>
qiangqiang199's avatar
qiangqiang199 committed
473

xuchao's avatar
xuchao committed
474
# Magic-doc
475

xuchao's avatar
xuchao committed
476
477
478
[Magic-Doc](https://github.com/InternLM/magic-doc) Fast speed ppt/pptx/doc/docx/pdf extraction tool

# Magic-html
479

xuchao's avatar
xuchao committed
480
481
482
483
484
485
486
[Magic-HTML](https://github.com/opendatalab/magic-html) Mixed web page extraction tool

# Links

- [LabelU (A Lightweight Multi-modal Data Annotation Tool)](https://github.com/opendatalab/labelU)
- [LabelLLM (An Open-source LLM Dialogue Annotation Platform)](https://github.com/opendatalab/LabelLLM)
- [PDF-Extract-Kit (A Comprehensive Toolkit for High-Quality PDF Content Extraction)](https://github.com/opendatalab/PDF-Extract-Kit)