README.md

metadata

title: QuickPDF OCR
emoji: 📄
colorFrom: blue
colorTo: blue
sdk: docker
app_port: 8000
pinned: false

QuickPDF Studio: OCR Microservice

A dedicated, self-hosted OCR extraction API built with FastAPI, EasyOCR, and PyMuPDF.

Features

• Pure REST API: No frontend bloat, just high-performance endpoints.
• Multi-Format Support: JPG, PNG, WebP, and multi-page scanned PDFs.
• Localized: Pre-configured for English ('en') and German ('de').
• Privacy First: Files are processed in memory and never persisted to disk.
• Docker Ready: Easy deployment on any system.

Quick Start (Local)

1. Install Dependencies:

   pip install -r requirements.txt
   

2. Run Server:

   uvicorn main:app --reload
   

3. Check Health:

   curl http://localhost:8000/health
   

API Usage

POST /ocr

Extract text from a document.

Parameters:

• file: Multipart/form-data upload.
• languages (Optional): Query param (e.g. ?languages=en,de).

Example Request (Curl):

curl -X 'POST' \
  'http://localhost:8000/ocr?languages=en,de' \
  -H 'accept: application/json' \
  -H 'Content-Type: multipart/form-data' \
  -F 'file=@sample.pdf'

Example Response:

{
  "success": true,
  "text": "Full extracted text...",
  "pages": [
    {"page": 1, "text": "Page 1 text..."}
  ],
  "language_used": ["en", "de"]
}

Docker Deployment

1. Build Image:

   docker build -t quickpdf-ocr .
   

2. Run Container:

   docker run -d -p 8000:8000 --name ocr-service quickpdf-ocr
   

Notes

• Models: On first run, EasyOCR will download approximately 150MB of trained models to ~/.EasyOCR.
• Resources: This service is optimized for CPU. It is recommended to have at least 2GB of RAM available.