🌿 AgriVision: AI Smart Agriculture

AgriVision is an AI-powered, fully client-side smart agriculture platform that helps farmers diagnose plant diseases and predict crop health risks — no backend required.

---

✨ Features

Feature	Description
🔬 Disease Diagnosis	Upload a leaf image → instant AI-powered disease detection with organic & chemical cure recommendations
📊 Risk Prediction	Input weather data (temp, humidity, rainfall) → predict disease risk before infection occurs
📍 Auto Weather	Use your device's GPS + OpenWeatherMap to auto-fill weather conditions
📋 Prediction History	All predictions saved locally — review and clear anytime
⚡ 100% Client-Side	Runs TFLite AI model directly in your browser via TensorFlow.js & WASM — no server needed
📱 Fully Responsive	Mobile-first design with hamburger sidebar for all screen sizes

---

🚀 Quick Start

Option 1: Use the Live Demo

👉 https://ai-smart-agriculture.vercel.app/

Option 2: Run Locally (Frontend Only)

git clone https://github.com/codeXsidd/AI-Smart-Agriculture.git
cd AI-Smart-Agriculture/frontend

# Serve over HTTP (required for model fetch to work)
npx serve .
# or
python -m http.server 8080

Then open http://localhost:8080 in your browser.

⚠️ Important: Opening index.html directly via file:// won't work because of browser fetch restrictions. Always serve via HTTP.

Option 3: Run with Backend (Advanced)

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

The FastAPI server will start at http://127.0.0.1:8000. You can test endpoints at http://127.0.0.1:8000/docs.

---

🧠 How It Works

User uploads leaf image
        ↓
TensorFlow.js loads TFLite model (~11 MB, runs in-browser via WASM)
        ↓
Image preprocessed: resized to 224×224, normalized [0,1]
        ↓
TFLite model runs inference → 65-class probability distribution
        ↓
Top class selected → Disease name + cure recommendation displayed

For Risk Prediction, a scikit-learn RandomForest model was pre-computed into a JavaScript lookup table (risk_model.js), enabling instant serverless predictions.

---

📁 Project Structure

AI-Smart-Agriculture/
├── frontend/
│   ├── index.html          # Disease Diagnosis page
│   ├── before.html         # Risk Prediction page  
│   ├── history.html        # Prediction History page
│   ├── style.css           # Premium glassmorphism UI
│   ├── script.js           # Client-side AI inference engine
│   ├── disease_data.js     # 65-class disease labels + cure database
│   ├── risk_model.js       # Exported sklearn RandomForest as JS
│   └── model/
│       └── smart_agri_model_quant.tflite   # Quantized TFLite model
└── backend/
    ├── main.py             # FastAPI server (optional)
    ├── cure_dict.py        # Disease cure dictionary
    ├── requirements.txt    # Python dependencies
    ├── model1/             # TFLite disease classification model
    └── model2/             # sklearn risk prediction model + encoders

---

🛠️ Tech Stack

Layer	Technology
Frontend	HTML5, Vanilla CSS (Glassmorphism), JavaScript ES2022
AI Runtime	TensorFlow.js 4.x + TFLite WASM Runtime
Disease Model	MobileNetV2-based TFLite (~11MB, 65-class PlantVillage)
Risk Model	RandomForest (scikit-learn, exported as JS)
Weather API	OpenWeatherMap API
Backend (Optional)	Python, FastAPI, Uvicorn
Deployment	Vercel (frontend), Render (backend API optional)

---

🌾 Supported Crops (Disease Diagnosis)

Apple • Blueberry • Cherry • Corn/Maize • Grape • Orange • Peach • Pepper • Potato • Raspberry • Soybean • Squash • Strawberry • Tomato

🌾 Supported Crops (Risk Prediction)

Potato • Rice • Wheat

---

🤝 Contributing

Contributions are welcome! Please feel free to:

• Submit bug reports or feature requests via Issues
• Fork the repo and submit a Pull Request
• Star ⭐ the repo if you find it useful!

---

📝 License

MIT License — free to use, modify, and distribute.