PaddleOCR

What it is

A Python toolkit and set of models for reading text from images and PDFs in more than 100 languages, including scene text, tables, formulas, and seals. Its document pipelines and a small vision-language model convert pages into structured Markdown or JSON with layout coordinates. Developers building RAG systems, data extraction jobs, or document translation use it as the parsing step, running it on CPU, GPU, or other accelerators, or serving it over an API.