ocr_kb

Extract text, LaTeX formulas, and figures from academic PDFs into per-page Markdown and DOCX.

247|14|Updated Feb 12, 2026

Search Tags:#ocr #pdf #academic #pipeline #checkpoint #docx #latex

Ready to run

npx skills add https://github.com/TFboy1/academic-paper-writer --skill ocr-kb

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: ocr_kb
Source: https://github.com/TFboy1/academic-paper-writer/tree/main/ocr_kb
Command: npx skills add https://github.com/TFboy1/academic-paper-writer --skill ocr-kb

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pandoc, PyMuPDF, and includes scripts (resource) components.

What problem does it solve?

通过多模态大模型对长文档的每一页进行扫描与处理，精确提取文本、LaTeX公式以及独立科研配图，并将中间产物放在 resources/，最终交付放在 outputs/。支持环境清理、断点恢复和全局编号管理，确保复杂论文工作流的可重复性与可追溯性。

Core Features & Use Cases

每页提取后生成页级 Markdown，便于后续排版与审稿。
提供公式与图片的全局编号、裁剪与嵌入，确保跨页面的一致性与可追踪性。
Use Case: 将带有大量公式与图表的长文档转化为可直接使用的 DOCX/Markdown 工作流，便于IEEE/ACM等模板排版与版本控制。

Quick Start

Process a multi-page PDF to extract text, LaTeX formulas, and figures, generating per-page Markdown and incremental DOCX with checkpointing.

Frequently Asked Questions about ocr_kb

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I convert long academic PDFs to editable DOCX with OCR?▼

To convert long academic PDFs to editable DOCX with OCR, automate end-to-end extraction of text, LaTeX formulas, and figures into per-page Markdown, then compile incremental DOCX. This workflow requires Pandoc and PyMuPDF to process and render complex research papers.

Does the PDF to DOCX OCR pipeline support LaTeX formula extraction?▼

Yes, the PDF to DOCX OCR pipeline supports LaTeX formula extraction by using Pandoc to convert LaTeX into OMML format. This ensures mathematical equations are accurately extracted and embedded in the final editable DOCX file.

How do I extract figures and text from research notes per-page?▼

You can extract figures and text from research notes per-page by rendering each PDF page with PyMuPDF and generating page-level Markdown. This process includes global numbering for figures to ensure cross-page consistency and traceability.

Can I resume OCR processing for long documents if the workflow stops?▼

Yes, you can resume OCR processing for long documents because the workflow supports checkpointing and breakpoint recovery. Intermediate products are saved in a strict directory structure, ensuring complex paper workflows remain repeatable and traceable.

Do I need Pandoc and PyMuPDF to extract text from academic PDFs?▼

Yes, you need Pandoc and PyMuPDF to extract text from academic PDFs in this workflow. PyMuPDF handles page rendering, while Pandoc is strictly required for converting extracted LaTeX formulas into OMML format for DOCX compatibility.