LinguaLearn System

A desktop application for capturing and preserving indigenous language terms using AI-powered object recognition.

Features

• Real-time object detection using Meta's Segment Anything Model (SAM)
• Voice recording for local term capture
• Support for South African languages
• Offline capability for areas with limited connectivity
• Visual dictionary of terms with object recognition

Prerequisites

• Node.js v16 or higher
• Python 3.8 or higher
• Webcam for object detection
• Microphone for voice recording

Installation

1. Clone the Repository

git clone https://github.com/ZubeidHendricks/lingualearn-system.git
cd lingualearn-system

2. Install Dependencies

There are known dependency conflicts that need to be resolved during installation. Follow these steps in order:

# Install react-scripts with legacy peer deps
npm install react-scripts --legacy-peer-deps

# Install required babel plugin
npm install @babel/plugin-proposal-private-property-in-object --legacy-peer-deps

# Install Python dependencies
pip install -r requirements.txt

3. Install Model Files

Download the required model files:

• Download the SAM model file from Meta
• Place it in the models directory
• Update the model path in config if necessary

Running the Application

# Start the application (this will start both React and Electron)
npm start

Common Issues and Solutions

React Scripts Not Found

If you encounter the error "'react-scripts' is not recognized", run:

npm install react-scripts --legacy-peer-deps

Dependency Conflicts

If you see dependency conflict errors, use the --legacy-peer-deps flag:

npm install --legacy-peer-deps

Development Setup

Environment Setup

1. Camera access is required for object detection
2. Microphone access is required for voice recording
3. Ensure Python environment has all ML dependencies installed

Directory Structure

lingualearn-system/
├── src/
│   ├── lingualearn/      # Python backend code
│   ├── main/            # Electron main process
│   └── components/      # React components
├── models/             # AI model files
├── tests/             # Test files
└── docs/              # Documentation

Contributing

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

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

Acknowledgements

• Meta's Segment Anything Model (SAM)
• OpenAI's Whisper for voice recognition
• The indigenous language communities of South Africa