Yuliang-Liu/MonkeyOCRv2

Health v2 · maintenance only

Flags: no_releases young no_license

Adoption not part of the score

1256 stars · 125 forks observed · 2026-09-03

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

MonkeyOCRv2 is a document-native vision encoder and visual-text foundation model for Document AI, pretrained on the 113M-image MonkeyDoc v2 corpus across 17 languages. It ships as a standalone vision backbone plus downstream models for document parsing, text/formula recognition, detection, and document understanding.

Use cases

• parse documents into markdown
• extract text from scanned pdfs
• recognize math formulas in images
• detect text in scene photos
• use a document vision encoder for downstream models
• run multilingual ocr on photographed documents
• detect document tampering

When to choose

• you need state-of-the-art open-source document parsing across many languages
• you want a document-native vision backbone to replace CLIP/DINO/SAM encoders
• you need fast inference via vLLM or CPU-only parsing

When to avoid

• you only need OCR on natural photos with sparse text
• you need a permissively licensed model (license is not standard)
• you want a general-purpose vision model for non-document imagery

Facets

library · maturity active

ocr machine-learning deep-learning image-processing llm-inference artificial-intelligence computer-vision pdf large-language-models python document-ai vision-encoder foundation-model document-parsing pretrained-model multilingual-ocr formula-recognition vllm natural-language-processing linux gpu docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Yuliang-Liu/MonkeyOCRv2")