CPU-only fast OCR for screenshots, images, PDFs, webpages

This project provides a fast, CPU-only OCR pipeline for extracting text from screenshots, images, PDFs, and webpages, making it ideal for developers and researchers who lack GPU access. It is interesting because it achieves near-GPU-level speed and accuracy using optimized models like TrOCR and PaddleOCR, enabling efficient local document processing without cloud dependencies.