📄 Baidu Introduces Unlimited-OCR
Baidu has released Unlimited-OCR, a high-performance model with 3 billion parameters designed for one-shot document analysis. It is optimized for long-horizon processing, allowing it to efficiently parse both individual images and multi-page PDF files in both "gundam" and "base" modes.
🌍 The shift to the "one-shot long-horizon parsing" concept significantly simplifies document processing pipelines, replacing complex chains of preprocessing and segmentation with a single universal model.
👤 The tool is useful for developers and researchers to automate data extraction from complex PDFs and scans without the need for manual preparation of every page.
Source 1: https://huggingface.co/baidu/Unlimited-OCR Source 2: https://github.com/baidu/Unlimited-OCR