Food Ingredient Risk Classifier

An AI-powered application that classifies food ingredients into risk categories and provides concise safety explanations. This tool helps consumers make informed decisions about the products they buy by analyzing ingredient lists and identifying potential health concerns.

Features

• Ingredient Risk Classification: Classifies food ingredients into 5 risk levels (1-5) using a fine-tuned DeBERTa model
• Concise Safety Explanations: Provides clear, concise explanations for each ingredient's safety profile
• Web Interface: User-friendly Streamlit interface for easy interaction
• API Access: RESTful API for programmatic access to the classification engine

Risk Levels

• 🟢 Level 1-2: Very Safe/Safe - Natural ingredients
• 🟡 Level 3: Moderate - Refined but generally safe
• 🟠 Level 4: Concerning - Artificial additives
• 🔴 Level 5: High Risk - Potentially harmful substances

Tech Stack

• Backend: FastAPI
• Frontend: Streamlit
• AI Models:
  • DeBERTa transformer for ingredient risk classification
  • Groq LLM API for generating explanations
• Vector Storage: FAISS
• Embeddings: Sentence Transformers

Prerequisites

• Python 3.8+
• Groq API key (set in .env file)

Installation

1. Clone the repository:

   git clone <repository-url>
   cd bt-el

2. Create a virtual environment:

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Set up environment variables: Create a .env file in the project root with your Groq API key:

   GROQ_API_KEY=your_api_key_here
   

Usage

Running the Backend Server

uvicorn backend:app --reload

The backend API will be available at http://localhost:8000.

Running the Frontend Application

streamlit run frontend.py

The web interface will be available at http://localhost:8501.

API Endpoints

• POST /predict - Classify ingredients and get risk assessment
• GET /health - Health check endpoint
• GET / - API information

Example API Request

curl -X POST http://localhost:8000/predict \
  -H "Content-Type: application/json" \
  -d '{"text": "refined wheat flour, sugar, edible vegetable oil (palmolein), emulsifier (322), synthetic food colour (INS 133)"}'

Project Structure

.
├── backend.py              # FastAPI backend server
├── frontend.py             # Streamlit web interface
├── rag_pipeline.py         # LLM explanation generation pipeline
├── dataset_documentation.md # Ingredient safety documentation
├── requirements.txt        # Python dependencies
└── .env                   # Environment variables (not included in repo)

Contributing

1. Fork the repository
2. Create a 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.