NewWeekly newsletter with catalog picks and agent news · Read
ToolsMCPBlogResearchCommunityStar on GitHub
All Tools
C
DataFreemiumOpen Source

CHANDRA

Layout-aware OCR for complex documents

Apache-2.0

ABOUT

Generic OCR drops checkboxes, messy tables, handwriting, and reading order, so downstream RAG and extraction jobs get unusable text. Chandra is a document OCR model that emits HTML, Markdown, or JSON with layout, and it handles forms, math, multilingual pages, and handwriting that simpler extractors miss.

INSTALL
pip install chandra-ocr

INTEGRATION GUIDE

1. Convert PDFs with complex tables into Markdown or HTML that keeps cell structure 2. Extract handwritten forms and checkboxes into structured JSON for downstream systems 3. Build a multilingual OCR pipeline covering 90-plus languages including math layouts 4. Run Chandra locally with vLLM or Hugging Face, or use the Datalab hosted API

TAGS

pythonocrpdfdocumentstableshandwritingdatalab
Chandra — AI Tool | Agentic AI For Good