ShelfieText

A specialized OCR system for processing book spines, built upon the capabilities of both EasyOCR and Tesseract. This project is an independent fork and adaptation of BetterOCR by Junho Yeo.

Features

• OCR Combination: Utilizes the power of both EasyOCR and Tesseract to extract text from images.
• Specialized for Book Spines: Optimized for the unique challenges posed by text on book spines.
• FastAPI Backend: Provides a robust and efficient backend server for processing OCR requests via a RESTful API.

Getting Started

Install packages:

pip install -r requirements.txt

Create a tessdata folder with the trained data file:

wget https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata -O ./tessdata/eng.traineddata

Running the server

After setting up, you can start the FastAPI server by running:

uvicorn main:app

Testing

The examples/ directory contains sample images that you can use to test the system.

Using curl:

curl -X 'POST' 'http://127.0.0.1:8000/ocr/' \
  -H 'accept: application/json' \
  -F 'image=@examples/sample.jpg' \
  -F 'lang=en'

License This project is licensed under the MIT License. See the LICENSE file for more details.