GreenGuard - An AI-driven Interactive Chatbot for Real-time Crop Disease Detection using AGDFNet

Prototype: https://youtu.be/OE9KS9Ulcos?si=G_Co71CIU6b5h9o2

Research Paper: https://docs.google.com/document/d/16dlcJQwMlwsY-dh49TIpj6FXRT9mQ0T65wFbP-Y_faE/edit?tab=t.0

Project Overview

GreenGuard is an AI-powered, interactive web application designed to assist farmers and agricultural specialists in detecting and identifying plant diseases using image-based diagnostics. By leveraging the AGDFNet (Attention-Guided Multi-Scale Disease Feature Network) deep learning model, the system provides real-time disease identification, treatment recommendations, and preventive insights, all integrated into an intuitive chatbot interface.

💡 Key Features

🧠 Real-time Plant Disease Detection

• Upload or capture leaf images for instant analysis
• Offline model inference using AGDFNet architecture
• Real-time feedback with disease probability scores
• Support for both online and offline modes

🌾 Multi-Crop, Multi-Disease Coverage

• Comprehensive coverage of 61 disease classes
• Support for 11 major crop types:
  • Bean
  • Cotton
  • Groundnut
  • Maize
  • Pepper Bell
  • Potato
  • Rice
  • Spinach
  • Sugarcane
  • Tomato
  • Turmeric

📊 Rich Disease Reports

• Class-specific precision, recall, and F1-scores
• Disease probability heatmaps
• Growth stage analysis
• Severity assessment
• Environmental factor correlation

💊 Treatment & Prevention

• Personalized treatment prescriptions
• Preventive measures and best practices
• Fertilizer recommendations
• Integrated pest management suggestions
• Seasonal care guidelines

💬 AI Chatbot Interface

• Interactive diagnosis conversations
• Treatment recommendations
• Fertilizer information
• Location-based mapping
• Weather integration
• Growth stage advice

🕒 Detection History

• Comprehensive scan history
• Progress tracking
• Treatment effectiveness analysis
• Exportable reports
• Seasonal comparison
• Growth pattern analysis

📱 Mobile Optimization

• Offline model inference
• Online treatment recommendations
• Camera integration
• Mobile-friendly UI
• Offline storage
• Network-optimized images

📚 Usage Guide

1. Home Page

• Upload or capture a crop leaf image
• Support for JPG, PNG, and JPEG formats
• Maximum file size: 5MB
• Real-time image preview

2. Detection Process

• Image processed through AGDFNet
• Multi-scale feature extraction
• Attention-based disease localization
• Real-time probability calculation

3. Report Generation

• Display of detected disease class
• Probability score visualization
• Treatment recommendations
• Prevention measures
• Export options

4. Chatbot Interaction

• Ask questions about diagnosis
• Get localized advice
• Treatment follow-up
• Prevention queries
• Growth stage information

5. History Management

• Access previous scans
• Compare results
• Track treatment progress
• Export historical data
• Analyze patterns

🛠 Troubleshooting

1. Upload Errors

• Error: Unsupported file format

  • Solution: Use JPG, PNG, or JPEG only
  • Tip: Convert images using free tools

• Error: File size too large

  • Solution: Compress image before upload
  • Tip: Use image compression tools

2. Backend Connection

• Error: Flask server not running
  • Solution: Start backend server
  • Command: python backend.py
  • Port: 5000

3. Model Loading

• Error: Missing model file
  • Solution: Place adgf_net.keras in models/
  • Location: backend/models/adgf_net.keras
  • Tip: Verify file permissions

4. Chatbot Issues

• Error: API key not found
  • Solution: Check .env file
  • Variable: GROQ_API_KEY
  • Tip: Get key from Groq Dashboard

🛠️ Technologies Used

Layer Stack

Frontend

• Framework: React 18+
• Build Tool: Vite
• Styling: TailwindCSS
• Animations: Framer Motion
• UI Components: Material UI
• State Management: React Context
• Image Processing: HTML5 Canvas

Backend

• Framework: Flask 2.x
• CORS: Flask-CORS
• ML Integration: TensorFlow 2.x
• Chat Integration: LangChain
• API Routing: Flask-RESTful
• Error Handling: Flask-Error

Machine Learning

• Model Architecture: AGDFNet
• Loss Function: Adaptive Focal Loss
• Attention Mechanisms: Spatial & Channel Attention
• Transfer Learning: Pre-trained CNN
• Optimization: AdamW
• Metrics: Precision, Recall, F1-Score

Storage & Data

• Session Management: SessionStorage
• Image Storage: Temporary filesystem
• History: IndexedDB
• Cache: Redis (optional)

API & Integration

• LLM Provider: Groq
• Weather API: OpenWeatherMap
• Geolocation: Geolocation API
• Image Processing: Pillow
• Database: SQLite (optional)

🚀 Setup Instructions

Prerequisites

• Node.js (v18+)
• Python (v3.9+)
• NPM or Yarn
• Internet connection for API calls
• Git
• Python virtual environment

Frontend Setup

# Clone the repository
$ git clone <repository-url>
$ cd GreenGuard

# Install dependencies
$ npm install

# Run development server
$ npm run dev

# Access the application at http://localhost:5173

Backend Setup

# Navigate to the backend directory
$ cd backend

# Create and activate virtual environment
$ python -m venv venv
# On Windows:
$ venv\Scripts\activate
# On macOS/Linux:
$ source venv/bin/activate

# Install dependencies
$ pip install -r requirements.txt

# Run the backend server
$ python backend.py

# The API will be available at http://localhost:5000

Environment Configuration

1. Create a .env file in the backend directory with:

GROQ_API_KEY=your_groq_api_key
OPENWEATHER_API_KEY=your_openweather_api_key

2. Get your API keys from:

• Groq Dashboard: https://console.groq.com
• OpenWeatherMap: https://openweathermap.org/api

Project Structure

GreenGuard/
├── backend/                  # Python Flask API
│   ├── models/               # ML model files
│   │   └── adgf_net.keras    # Custom CNN model
│   ├── temp_uploads/         # Temporary storage for uploaded images
│   ├── backend.py            # Main Flask application
│   ├── disease_predict.py    # ML model implementation
│   ├── groq_demo.py          # Integration with Groq API
│   └── requirements.txt      # Python dependencies
│
├── src/                      # React frontend
│   ├── assets/               # Static assets
│   ├── components/           # Reusable UI components
│   ├── context/              # React context providers
│   ├── hooks/                # Custom React hooks
│   ├── pages/                # Application pages
│   ├── App.jsx               # Main application component
│   ├── App.css               # Application styles
│   ├── index.css             # Global styles
│   └── main.jsx              # Entry point
│
├── public/                   # Public static files
├── .gitignore                # Git ignore rules
├── package.json              # NPM dependencies
├── tailwind.config.js        # Tailwind CSS configuration
└── vite.config.js            # Vite configuration