Verified project record

PaddlePaddle/PaddleOCR

PaddleOCR converts PDF documents, images, and Office files into structured formats including JSON, Markdown, and DOCX. The project features PaddleOCR-VL-1.6, a 0.9B parameter vision-language model, alongside PP-StructureV3 layout conversion.

Project overview

Provides document parsing using a lightweight vision-language model and structured output generation for retrieval and agentic pipelines.

Project type: RAG · Document Processing · Infrastructure
Use cases: Documents & Office · Knowledge Q&A
Deployment: Refer to project documentation
License: Apache-2.0

Best for

Developers and AI engineers building document OCR processing and retrieval pipelines.

Key capabilities

Features PaddleOCR-VL-1.6, a 0.9B parameter lightweight vision-language model for document parsing that outputs structured Markdown and JSON.
Converts complex PDFs and images into Markdown or JSON, providing fine-grained coordinate information including table cell and text coordinates.
Converts common document formats such as Word, Excel, and PowerPoint into Markdown.
Exports parsed results to DOCX for viewing and editing in Microsoft Word.
Provides an official browser inference SDK that supports running PP-OCRv5 directly in the browser.
Converts models to ONNX format for deployment flexibility.
Accelerates inference using engines like OpenVINO, ONNX Runtime, and TensorRT.
Accelerates inference using multi-GPU and multi-process configurations.

Limitations and risks

Full functionality is dependent on Paddle or supported inference backends.

Getting started

Install via pip, refer to PP-OCR or PP-Structure documentation, and run the pipeline locally using a Python environment.

Evidence and sources

GitHub project description: 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.
README: SOTA Document VLM: Featuring PaddleOCR-VL-1.6 (0.9B), the industry's leading lightweight vision-language model for document parsing. It achieves 96.3% accuracy on OmniDocBench v1.…
README: 100+ Languages Supported: Native recognition for a vast global library. PP-OCRv6 supports 50 languages with a single unified model (Chinese, English, Japanese, and 46 Latin-script…
README: ![hardware](https://img.shields.io/badge/hardware-cpu%2C%20gpu%2C%20xpu%2C%20npu-yellow.svg)
README: ![python](https://img.shields.io/badge/python-3.8~3.12-aff.svg) ![os](https://img.shields.io/badge/os-linux%2C%20win%2C%20mac-pink.svg)

Reviewed at: 2026-07-17T04:36:35Z