---
id: paddleocr
version: "3.7.0"
license: Apache License 2.0
license_treatment: permissive
maintenance: active
---

# paddleocr — Awesome multilingual OCR and document parsing toolkits based on PaddlePaddle

License: permissive · Maintenance: active · Downloads: 3.2M/mo

## What it is and what it does

PaddleOCR is a multilingual optical character recognition and document parsing toolkit that converts images and PDFs into structured, machine-readable formats (JSON or Markdown). It combines traditional scene text recognition with specialized vision-language models for document understanding, supporting over 100 languages through unified models that eliminate the need for language switching. The package includes multiple model tiers (tiny, small, medium) optimized for different deployment scenarios—from edge devices to cloud servers—and handles complex document elements like tables, formulas, and charts alongside plain text.

The toolkit is designed for building RAG and agentic AI applications, with integration points for popular frameworks. Runtime dependencies are minimal: PyYAML, requests, aiohttp, typing-extensions, and paddlex, keeping the installation footprint light despite the capability breadth.

Use it for:

- Extract text and tables from scanned documents or PDFs for data entry automation or archival systems.
- Convert business documents (invoices, receipts, forms) into structured JSON for downstream processing or database ingestion.
- Build document-aware RAG pipelines by parsing PDFs into Markdown and feeding them to LLM retrieval systems.
- Recognize text in natural scene images (street signs, license plates, industrial labels) for computer vision applications.
- Parse multilingual documents without model switching, handling mixed-language content in a single pass.
- Deploy lightweight OCR on edge devices or mobile environments using the tiny model tier.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

PaddleOCR extracts text, tables, and structured data from images and PDFs, converting them into JSON or Markdown format with support for over 100 languages and specialized document parsing models.

Yes. PaddleOCR is actively maintained, permissively licensed, has low install friction, and zero known vulnerabilities. It addresses a well-defined problem (document parsing and multilingual OCR) with broad ecosystem integration. Install it if you need OCR or document-to-structured-data conversion; the main gotcha is that models download on first use.

## Install

pip install paddleocr
uv add paddleocr
poetry add paddleocr

## Installing paddleocr

Before you install: Low friction installation with a pure Python wheel. Active maintenance with recent releases and strong community signal. Supports Python 3.8–3.13.

License in practice: Apache License 2.0 is permissive, allowing commercial and private use with minimal restrictions—suitable for most production deployments.

Quickstart:

pip install paddleocr

from paddleocr import PaddleOCR
ocr = PaddleOCR(use_angle_cls=True, lang='en')
result = ocr.ocr('image.jpg', cls=True)

Models are downloaded on first use; initial run may require network access and disk space for model files.

Verify before relying:

- Exact model download size and disk space requirements for different model tiers (tiny, small, medium).
- Performance benchmarks on specific hardware (GPU models, CPU types) beyond the general speedup claims.
- Whether all 100+ languages are equally accurate or if accuracy varies significantly by language.
- Real-world latency and throughput on production-scale document batches.

## Package facts

- License: Apache License 2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 3.2M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

ocr text extraction, pdf to markdown conversion, multilingual document parsing, table recognition from images, document structure extraction, vision language model ocr, pdf document ai, structured data from images, document-parsing, multilingual, vision-language-model

[View on SkillFed](https://skillfed.io/packages/paddleocr) · [View on PyPI](https://pypi.org/project/paddleocr/)