FreshCam

A smart fruit freshness detection app that uses computer vision and AI to identify fruits and determine their ripeness.

Features

• 🍎 Fruit Identification: Uses Google Gemini AI to accurately identify fruit types
• 🎯 Ripeness Detection: Computer vision model detects if fruit is unripe, ripe, or overripe
• 🔄 AI Fallback: Automatically uses Gemini AI as fallback when CV model fails, ensuring reliable results
• 📱 Mobile App: React Native/Expo frontend for easy photo capture
• ⚡ FastAPI Backend: High-performance API for image analysis

Setup

Backend Setup

1. Navigate to the backend directory:

cd backend

2. Install Python dependencies:

pip install -r requirements.txt

3. Create a .env file with your API keys:

ROBOFLOW_URL=https://detect.roboflow.com
ROBOFLOW_API_KEY=your_roboflow_api_key
GEMINI_API_KEY=your_gemini_api_key

Get your API keys:

• Roboflow: https://roboflow.com
• Google Gemini: https://makersuite.google.com/app/apikey

4. Run the backend:

python app.py

Frontend Setup

1. Navigate to the frontend directory:

cd frontend

2. Install dependencies:

npm install

3. Run the app:

npm start

4. Start the app:

npx expo start --tunnel

How It Works

1. Take a photo of a fruit using the mobile app
2. CV Model analyzes the image for ripeness
3. Gemini AI identifies the fruit name (always)
4. Fallback mechanism: If CV model fails, Gemini AI analyzes ripeness
5. Results returned: fruit name, ripeness stage, and confidence score

API Response Format

{
  "fruit_name": "apple",
  "ripeness": "ripe",
  "confidence": 92.5,
  "source": "cv_model"
}

• source can be: "cv_model" or "gemini_fallback"