Open-source text recognition and document parsing for images and PDFs
Overview
- Purpose
- Open-source text recognition and document parsing for images and PDFs
- Category
- Office
- Developer
- PaddlePaddle
- Official site
- https://github.com/PaddlePaddle/PaddleOCR (opens in a new tab)
Main features
- Recognizes text in images and scanned documents
- Layout and structure parsing for documents
- Some models recognize tables and formulas and output Markdown or JSON
- Deploy and call recognition models in your own projects
Best for
Suited to developers and data staff processing scans, receipts or documents in bulk who are willing to set up a recognition environment.
Things to know
- Capabilities differ by model and installation; choose according to the official documentation
- Recognition of text, tables and formulas can still be wrong; check important data against the original
FAQ
- Is it a web page you can just open and use?
- The entry here is the open-source project, which usually needs installation and configuration or integration into an existing app.
- Can it output structured documents?
- Yes. The official project states some document parsing models support Markdown and JSON output.