PaddleOCR-API

A CPU-only OCR service built with FastAPI and PaddleOCR for text and table extraction from PDF, PNG, and JPG documents.

Features

• Text Recognition: Extract text from documents using PaddleOCR with Russian language support
• Table Detection: Extract structured tables using PP-Structure
• Multiple Formats: Support for PDF, PNG, and JPG files
• Flexible Page Selection: Process specific pages with syntax like "1-3,5"
• Configurable DPI: Adjust PDF conversion quality
• HTML Output: Optional HTML representation of extracted content
• Health Check: Monitoring endpoint for service health

API Endpoints

GET /health

Health check endpoint that returns service status.

Response:

{
  "status": "healthy",
  "service": "PaddleOCR-API"
}

POST /analyze

Analyze documents for text and table extraction.

Parameters:

• file (required): Document file (PDF, PNG, or JPG)
• dpi (optional, default: 350): DPI for PDF conversion
• pages (optional, default: all): Page specification like "1-3,5"
• return_html (optional, default: false): Include HTML representation

Response:

{
  "lang": "ru",
  "pages": [
    {
      "index": 1,
      "text": {
        "full": "Extracted text content..."
      },
      "tables": [
        {
          "rows": [
            {"html": "<table>...</table>"}
          ],
          "bbox": [x1, y1, x2, y2]
        }
      ],
      "html": "<div class='page'>...</div>"
    }
  ]
}

Installation & Usage

Quick Start with Prebuilt Docker Image (Recommended)

No need to build locally! Use the ready-to-go image:

docker run -p 8000:8000 bvdcode/paddleocrapi:v1.0

The service will be available at http://localhost:8000

Using Docker Compose

1. Clone the repository (optional, only if you want to use docker-compose):

git clone https://github.com/bvdcode/PaddleOCR-API.git
cd PaddleOCR-API

2. Edit docker-compose.yml to use the prebuilt image:

services:
  paddleocr-api:
    image: bvdcode/paddleocrapi:v1.0
    ports:
      - "8000:8000"

3. Start the service:

docker-compose up

---

Local Build (for development)

1. Build the image manually:

docker build -t paddleocr-api .

2. Run the container:

docker run -p 8000:8000 paddleocr-api

Local Development

1. Install Python 3.12 and system dependencies:

# Ubuntu/Debian
sudo apt-get update
sudo apt-get install poppler-utils libgl1

# macOS
brew install poppler

2. Install Python dependencies:

pip install -r requirements.txt

3. Run the application:

python main.py

Or using uvicorn directly:

uvicorn main:app --host 0.0.0.0 --port 8000 --reload

Dependencies

• FastAPI: Modern web framework for building APIs
• Uvicorn: ASGI server for FastAPI
• PaddleOCR: OCR toolkit with Russian language support
• PP-Structure: Table detection and structure analysis
• pdf2image: PDF to image conversion
• Pillow: Image processing library
• python-multipart: File upload support

Usage Examples

Health Check

curl http://localhost:8000/health

Analyze PDF Document

curl -X POST "http://localhost:8000/analyze" \
  -F "file=@document.pdf" \
  -F "dpi=350" \
  -F "pages=1-3,5" \
  -F "return_html=true"

Analyze Image

curl -X POST "http://localhost:8000/analyze" \
  -F "file=@image.png"

Configuration

The service can be configured through environment variables:

• PYTHONUNBUFFERED=1: Ensure Python output is not buffered
• PYTHONIOENCODING=utf-8: Set UTF-8 encoding for Python I/O

System Requirements

• CPU: Multi-core processor recommended
• Memory: Minimum 2GB RAM, 4GB+ recommended
• Storage: At least 2GB free space for models and temporary files
• OS: Linux, macOS, or Windows with Docker

Performance Notes

• First request may take longer as PaddleOCR downloads required models
• PDF processing time depends on document size and page count
• CPU-only processing is slower than GPU but more compatible
• Consider using page selection for large documents to improve performance

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Support

For issues and questions, please create an issue in the GitHub repository.