ocr-benchmark

An on-premises, OCR-free unstructured data extraction, markdown conversion and benchmarking toolkit. (https://idp-leaderboard.org/)

Failure-oriented benchmark for PDF-to-Markdown, OCR, and VLM document parsers with executable assertions and public release leaderboards.

GlotOCR Benchmark

Reports, datasets, and experiments for DeepSolo-inspired Korean OCR reading-order correction.

Read scanned Japanese documents with several local OCR engines, each in its own environment

A one-stop-shop framework for integrating, benchmarking, and comparing OCR engines. Comes with configurable image preprocessing methods and build in evaluation pipeline. Supports PaddleOCR, EasyOCR and Tesseract out of the box.