diegomarzaa/pdf-ocr-obsidian

What it solves

It automates the conversion of PDF documents into structured Markdown files, specifically optimized for use in Obsidian. It solves the problem of manually extracting text and images from scanned PDFs while maintaining document hierarchy and image references.

How it works

The tool uses the Mistral AI OCR API to process PDFs. It scans a designated input folder for files, uploads them to the API, and then parses the response to create a Markdown file. Images are extracted separately and linked within the text using Obsidian-style wikilinks (![[image-name]]). To prevent redundant API calls, the system caches the raw OCR response as a JSON file.

Who it’s for

Users of Obsidian who want to import scanned PDF content into their knowledge base while preserving images and structure.

Highlights

• Batch Processing: Automatically processes multiple PDFs in a folder.
• Obsidian Integration: Uses wikilinks for images to ensure compatibility with Obsidian vaults.
• Flexible Interfaces: Available as both a local web application and a Jupyter Notebook.
• OCR Caching: Saves API responses to JSON to avoid repeated costs or calls.