Hybrid Crop Advisory System with AI-based Disease Detection

This project is a full-stack crop support platform that brings together:

• Rule-based crop sowing advisory
• Weather and forecast-based decision support
• AI-based disease detection
• User authentication
• NDVI-based field monitoring with Sentinel Hub

The goal is to help farmers and users make better crop decisions using weather conditions, field health insights, and image-based disease checks in one place.

---

Crop Advisory Module

The crop advisory module does not rely on machine learning.
Instead, it uses agronomic rules derived from practical agricultural conditions to generate recommendations.

Inputs Used

The advisory flow considers real farming parameters such as:

• Season (Kharif, Rabi, Zaid)
• Soil type
• Irrigation type (Rainfed / Irrigated)
• Weather conditions
• Forecast-driven suitability signals

Decision Logic

The system checks whether sowing is safe, risky, or should be delayed using threshold-based agricultural rules.

Examples of advisory logic:

• Extreme temperature can delay sowing
• Low rainfall with rainfed farming increases risk
• Continuous hot conditions can trigger heat-stress warnings
• Better moisture and weather suitability can support sow-now guidance

Output

The module produces structured results instead of a basic yes/no response:

• Sowing decision
• Risk level
• Crop recommendation or suitability guidance
• Action plan
• Reasoning behind the recommendation

Why Rule-Based Instead of ML?

Agriculture decisions often depend on established domain knowledge.
Using deterministic agronomic rules helps keep the advisory:

• Explainable
• Stable
• Consistent
• Less dependent on large historical training datasets

This makes the output easier to understand and trust.

---

AI Disease Detection

The project also includes an AI-based disease detection flow for crop images.

This part of the system is used to:

• Upload crop images
• Run disease detection
• Help users identify plant health issues more quickly

Disease Detection Model Link

https://drive.google.com/file/d/1EUVq695L63d1wbecpNyHkgBZwsEVTlkd/view?usp=sharing

---

NDVI Monitoring

The project includes an NDVI monitoring feature powered by Sentinel Hub for satellite-based field health analysis.

Implemented NDVI Features

• Protected NDVI pages in the frontend
• NDVI image generation for selected field bounds
• NDVI statistics for selected field geometry
• Health summary from NDVI values
• Field insights such as:
  • coverage percentage
  • uniformity score
  • weak and critical area estimates
  • urgency level
• Save field boundaries for authenticated users
• Save NDVI scan history
• View saved fields and field history

NDVI Backend Route

The backend exposes NDVI APIs under:

• /api/ndvi

Sentinel Hub Usage

The backend uses Sentinel Hub for:

• access token generation
• NDVI image processing
• NDVI statistics aggregation

---

Authentication and User Flow

The application includes authentication features for protected access to user-specific tools.

Implemented authentication features:

• Sign up
• Sign in
• JWT-based authentication
• Protected routes for dashboard and NDVI features
• User profile access through backend auth middleware

---

Main Frontend Routes

The project currently includes these main frontend flows:

• Language selection
• Home page
• Sign in / Sign up
• Dashboard
• Crop recommendation
• Weather
• Forecast
• Final recommendation
• Disease detection
• NDVI
• NDVI map

---

Backend API Areas

The backend currently serves routes for:

• /api/user
• /api/cropInput
• /api/weather
• /api/recommendation
• /api/ndvi

---

Tech Stack

• React
• React Router
• Node.js
• Express
• MongoDB
• JWT authentication
• Sentinel Hub APIs

---

Environment Variables

Frontend

REACT_APP_API_BASE_URL=https://your-backend-service.onrender.com
REACT_APP_ML_API_URL=https://your-ml-service.onrender.com
REACT_APP_FIREBASE_API_KEY=your-firebase-api-key
REACT_APP_FIREBASE_AUTH_DOMAIN=your-project.firebaseapp.com
REACT_APP_FIREBASE_PROJECT_ID=your-project-id
REACT_APP_FIREBASE_STORAGE_BUCKET=your-project.firebasestorage.app
REACT_APP_FIREBASE_MESSAGING_SENDER_ID=your-messaging-sender-id
REACT_APP_FIREBASE_APP_ID=your-firebase-app-id

Backend

PORT=5002
MONGODB_URI=your-mongodb-uri
JWT_SECRET=replace-with-a-long-random-secret
CORS_ORIGIN=https://your-frontend-domain.vercel.app
WEATHER_API_KEY=your-weather-api-key
CLIENT_ID=your-sentinel-client-id
CLIENT_SECRET=your-sentinel-client-secret

The NDVI service also supports these alternative Sentinel variable names:

SENTINEL_CLIENT_ID=your-sentinel-client-id
SENTINEL_CLIENT_SECRET=your-sentinel-client-secret

---

Role in the Overall System

The project combines multiple support layers:

1. Weather analysis for environmental awareness
2. Crop advisory for sowing decisions
3. Disease detection for crop health checks
4. NDVI-based monitoring for satellite field analysis

Together, these features form a more complete crop decision support platform.