pdf-craft – An open-source tool for converting PDF to Markdown

What is pdf-craft?

pdf-craft is a tool designed to convert PDF files into other formats (such as Markdown and EPUB), with a focus on processing scanned book PDFs. It supports extracting the main body content while filtering out non-body elements such as headers, footers, and footnotes. By combining the DocLayout-YOLO algorithm with PaddleOCR text recognition technology, pdf-craft can effectively handle cross-page issues and generate semantically coherent text.

The main functions of pdf-craft

• PDF to Markdown Function: Convert PDF files into Markdown format, extract the main text content while preserving the structure, and embed illustrations, tables, and formulas in the form of screenshots to ensure the generated Markdown file is semantically coherent.
• PDF to EPUB Function: Utilize large language models to construct the book structure of EPUB, generate a table of contents, integrate annotations and citations, correct OCR errors, and convert the content into EPUB format optimized for e-book readers.

The Technical Principles of PDF-Craft

• Page Layout Analysis: Perform layout analysis on PDF pages using the DocLayout-YOLO algorithm to identify the positions and boundaries of elements such as text blocks, images, and tables. Further optimize the layout parsing by combining a custom algorithm to ensure that the extracted main content is accurate and complete.
• Text Recognition: Conduct text recognition using PaddleOCR. PaddleOCR is a high-performance open-source OCR tool capable of accurately recognizing text content in scanned books. Pre-trained models are used to identify and extract text blocks from the page.
• Cross-Page Processing: When handling cross-page text, determine the logical relationships between text blocks based on algorithms to ensure the coherence of cross-page text.
• Reading Order Optimization: Use layoutreader to determine the reading order of text blocks. Generate a sequence that conforms to human reading habits based on the page layout and the positions of text blocks.

The project address of pdf-craft

• GitHub Repository: https://github.com/oomol-lab/pdf-

Application scenarios of pdf-craft

• Academic Research: Convert scanned academic papers into Markdown or EPUB format for easier editing, annotating, and organizing.
• E-book Production: Convert scanned books into EPUB format, generating tables of contents and chapter structures for convenient publishing and reading.
• Document Archiving: Convert paper documents or PDF files into Markdown or EPUB format for long-term archiving and retrieval.
• Educational Material Organization: Convert scanned textbooks or lecture notes into editable formats for convenient teacher organization and student learning.
• Personal Learning: Convert scanned books or materials into Markdown format for easy note-taking and review.