🌾 AgriVision AI - Smart Crop & Fertilizer Recommendation System

"Farming with Intelligence"

AgriVision AI is an advanced machine learning project designed to assist farmers in making data-driven decisions. By analyzing soil nutrients (Nitrogen, Phosphorous, Potassium) and environmental parameters (Temperature, Humidity, pH, Rainfall), the system predicts the most profitable crop to grow and suggests the optimal fertilizer to maximize yield.

🚀 Unique Features

1. 🧠 Smart Condition Analysis

Unlike standard prediction models that just give a label, AgriVision AI explains why.

• "High Nitrogen levels detected. Suitable for leafy growth."
• "Heavy rainfall region suitable for water-intensive crops."

2. 🛡️ Robust Input Validation

The system includes a creative "Sanity Check" layer that rejects illogical inputs with scientific context.

• Example: If you enter Temperature: 100°C, it replies: "🌡️ Temperature 100°C is too extreme for agriculture! Most crops die above 50°C."

3. 🧠 Deep Tech Dive: AI & ML Terminology

This project is built on robust Supervised Machine Learning principles. Here's a breakdown of the core technologies:

A. Random Forest Classifier (The Brain)

We use the Random Forest algorithm, which is an ensemble learning method.

• Ensemble Learning: Instead of relying on one decision tree, we create a "forest" of multiple trees.
• Voting Mechanism: Each tree gives a prediction (vote), and the class with the most votes becomes the model's output.
• Why it works: It reduces overfitting (memorizing data) and improves accuracy compared to a single Decision Tree.

B. Feature Engineering

The model doesn't just "guess"; it mathematically correlates input features to outputs.

• Inputs (X): N, P, K, Temperature, Humidity, pH, Rainfall.
• Target (Y): Crop Label (e.g., Rice, Maize).
• Label Encoding: Converting text labels (e.g., "Rice") into numbers (e.g., 0) so the machine can understand them.

C. Data Preprocessing

• Synthetic Data Generation: We created a custom dataset algorithm to simulate realistic agricultural patterns (e.g., Rice requiring high rain, Cotton requiring black soil).
• Normalization: Ensuring all numerical values are on a comparable scale.

D. Model Evaluation Metrics

• Accuracy Score: The ratio of correctly predicted observations to the total observations.
  • Crop Model: 99.77%
  • Fertilizer Model: 96.00%
• Confusion Matrix: Used during training to verify that the model doesn't confuse similar crops (e.g., Muskmelon vs. Watermelon).

E. Input Validation (Sanity Check Layer)

A rule-based logic layer that sits before the ML model.

• Physical Constraints: $0 \le Humidity \le 100$.
• Chemical Constraints: $0 \le pH \le 14$.
• Biological Limits: Checks for toxic nutrient levels (e.g., Nitrogen > 200).

graph TD
    A[User Input] --> B{Validation Layer}
    B -- Invalid --> C[Return Error]
    B -- Valid --> D[Feature Scaling]
    D --> E[Random Forest Model]
    E --> F[Prediction Label]
    F --> G[Dynamic Insight Generator]
    G --> H[Final Response]

Loading

4. 📊 High-Accuracy Models

The web interface is built with a translucent, futuristic design system, featuring animated backgrounds and loading states for a premium user experience.

🛠️ Tech Stack

• Frontend: HTML5, CSS3 (Glassmorphism), JavaScript (Fetch API)
• Backend: Python, Flask
• Machine Learning: Scikit-Learn (Random Forest), Pandas, NumPy
• Data processing: Custom synthetic data generation

⚙️ How to Run

1. Clone the repository (if you haven't yet).
2. Install dependencies:

   pip install -r requirements.txt

3. Run the Application:

   python3 app.py

4. Open http://127.0.0.1:5002 in your browser.

📂 Project Structure

AgriVision-AI/
├── app.py                # Flask Backend & ML Logic
├── data_generator.py    # Synthetic Data Creation
├── train_models.py      # Model Training Script
├── models/              # Saved .pkl models
├── static/              # CSS & JS
├── templates/           # HTML Templates
└── requirements.txt     # Dependencies

📈 Future Scope

• Integration with live weather APIs.
• Mobile App using React Native.
• Multi-language support for regional farmers.

---

Built with ❤️ for the farming community.