paddlepaddle/paddleocr
Turn any PDF or image document into structured data for your AI. A powerful, lightweight OCR toolkit that bridges the gap between images/PDFs and LLMs. Supports 100+ languages.
How paddleocr is put together
PaddleOCR is a multilingual OCR and document-parsing toolkit built on PaddlePaddle/PaddleX, published as the 'paddleocr' PyPI package with a CLI entry point (paddleocr.__main__). It is a polyglot monorepo: the core Python library delegates heavy lifting to the paddlex dependency, while sibling sub-packages provide a LangChain integration (langchain-paddleocr), an MCP server (mcp_server), JavaScript/TypeScript bindings (paddleocr-js), and official cloud-API SDKs in Go and TypeScript (api_sdk). Training/inference tooling lives under tools/ with YAML model configs under configs/, legacy PP-Structure pipelines under ppstructure/, and hardware deployment assets under deploy/. Data flow is typically image/PDF input -> OCR/document-parsing pipeline -> structured Markdown/JSON output, either locally via Paddle models or remotely via the token-authenticated PaddleOCR cloud API.
Languages
Frameworks
Infrastructure
Major components
paddleocr core package
Main Python library and CLI ('paddleocr' command) exposing OCR, document parsing, IE, and translation pipelines on top of paddlex.
api_sdk/go
Go client SDK for the PaddleOCR official cloud API with token auth, async job submission/polling, typed results, and resource download.
api_sdk/typescript
TypeScript client SDK for the same official PaddleOCR API, built with tsup.
mcp_server
Model Context Protocol server package (paddleocr_mcp) exposing PaddleOCR document parsing to MCP clients, with local-inference extras.
langchain-paddleocr
LangChain integration package wrapping PaddleOCR as LangChain-compatible components with strict typing/lint tooling.
paddleocr-js
npm workspace containing @paddleocr/paddleocr-js packages and a Vite demo app for browser/Node OCR usage.
tools/ + configs/
Training, evaluation, export, and inference scripts (train.py, eval.py, infer_*.py) driven by per-model YAML configs for det/rec/cls/table/sr tasks.
ppstructure + deploy/
Legacy PP-Structure prediction system plus deployment artifacts for C++ inference, Android, hubserving, and embedded ARM targets.
PaddleOCR gained a new document-parsing feature alongside a documentation refresh.
This week's activity on PaddleOCR focused on expanding the toolkit with a new parsing capability and refreshing its documentation. The team added support for HPD-Parsing, a new document-parsing feature, while also updating the README to keep the project's front page current.
Feature
HPD-Parsing support added
The project now supports HPD-Parsing, giving users a new way to parse documents within the OCR toolkit.
Docs
README updated
The main README file was refreshed to keep the project's introduction and instructions up to date.
Dependencies and code review
Dependency advisories
Security Watch
No known advisories across 0 scanned dependencies.
No known advisories in the scanned dependencies.
Code review
No concrete code-level issues confirmed.
Get this report every week for your repos.
GitZoid learns each repo, reports what changed, and flags what needs attention. One flat price for the whole team.
$19 a month, flat · First 10 outputs free · No card required