Visiting Card OCR and Text-to-Speech App

This application utilizes computer vision techniques to detect visiting cards from a live webcam feed, extract text from them using Azure Cognitive Services OCR (Optical Character Recognition), and then convert the extracted text to speech.

Features

• Real-time Visiting Card Detection: The application detects visiting cards in real-time from a live webcam feed using the YOLO (You Only Look Once) object detection model.

• Text Extraction: After detecting a visiting card, the application extracts the text from it using Azure Cognitive Services OCR API.

• Text-to-Speech Conversion: Once the text is extracted, the application converts it to speech using Azure Cognitive Services Text-to-Speech API.

• Quality-based Frame Selection: The application evaluates the quality of each frame to select the best frame for OCR processing, ensuring accurate text extraction.

Requirements

• Python 3.6 or higher
• OpenCV
• aiohttp
• azure-cognitiveservices-speech
• ultralytics

Installation

1. Clone the repository:

   git clone https://github.com/your-username/visiting-card-ocr.git

2. Install dependencies:

   pip install -r requirements.txt

3. Set up Azure Cognitive Services:

   • Sign up for an Azure account if you don't have one already.

   • Obtain the subscription key and region for Azure Cognitive Services Speech and Vision APIs.

   • Set these values in the .env file:

     COG_SERVICE_ENDPOINT=<your_cognitive_service_endpoint>
     COG_SERVICE_KEY=<your_cognitive_service_key>
     SPEECH_REGION=<speech_region>
     

Usage

1. Run the application:

   python main.py

2. Point your webcam towards visiting cards to detect and extract text from them.

3. Press 'q' to quit the application.

Notes

• Adjust the confidence threshold (threshold variable in main.py) for object detection as needed.
• Ensure that the webcam is properly connected and configured before running the application.

Credits

• YOLO (You Only Look Once) object detection model is implemented using the ultralytics library.
• Azure Cognitive Services is used for OCR (Optical Character Recognition) and Text-to-Speech conversion.
• Roboflow

License

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