Moduler

Uses OpenDataLoader PDF (fast local mode) for structured layout analysis. Extracts headings, paragraphs, tables, lists with bounding boxes. Requires Java 11+.

Uses marker-pdf for deep-learning-based PDF conversion. Accurate layout detection, table recognition, OCR, and equation handling. Requires Python 3.10+ and marker-pdf.

Uses MinerU CLI for structured PDF extraction — layout analysis, formula-to-LaTeX, table-to-HTML, OCR. Requires Python 3.10+ and mineru installed.

Uses IBM Docling for document conversion with advanced layout analysis, table extraction, and OCR. Requires Python 3.10+ and docling (pip install docling).