Maize_Classification-Model

---

Maize Disease Detection Web Application

Project Description

This project is a web application designed to assist in identifying and managing maize leaf diseases through AI-driven predictions. The application is dockerized for easy deployment and provides the following features:

1. Disease Prediction

   • Users can upload maize leaf images, and the system predicts the disease affecting the leaf using a pre-trained machine learning model.

2. Model Retraining

   • Users can upload bulk data (ZIP files of categorized images) to retrain the existing model with their dataset. This ensures the model adapts to new data and remains accurate.

3. Data Visualizations

   • A dedicated page displays various interactive visualizations based on the dataset, with key insights and interpretations.

4. Model Download

   • Users can download the retrained model for their use.

5. Evaluation Metrics

   • Detailed metrics for both prediction and retraining processes are displayed.

6. Performance Testing

   • A flood request simulation was conducted using Locust software to evaluate system performance under heavy load.

The application is built using HTML, CSS, and JavaScript for the frontend, while the backend is developed with FastAPI. The machine learning pipeline was designed from scratch, including scaling and monitoring on a cloud platform.

---

Demo Video

Watch a detailed demonstration of the project in action:
YouTube Demo

Live Links

Live Application

Access the live application here:
Application URL

Live Api Endpoint

Access the live API Endpoint here:
Endpoint URL

---

Setup Instructions

1. Clone the Repository

git clone https://github.com/yourusername/maize-disease-detection.git
cd maize-disease-detection

2. Set Up Docker

Ensure you have Docker installed. Build and run the Docker container:

# Build the Docker image
docker build -t maize-disease-detection .

# Run the container
docker run -p 8000:8000 maize-disease-detection

The application will be accessible at http://127.0.0.1:8000.

3. Set Up Without Docker (Optional)

If you prefer to run the application locally without Docker:

Install Dependencies

# Create a virtual environment
python -m venv venv

# Activate the virtual environment
# On Windows
venv\Scripts\activate
# On macOS/Linux
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

Run the Application

uvicorn main:app --reload

Access the application at http://127.0.0.1:8000.

---

Features Overview

1. Disease Prediction

• Endpoint: /predict
• Input: Single image upload.
• Output: Predicted disease class and confidence score.

2. Model Retraining

• Endpoint: /retrain/upload
• Input: ZIP file of categorized images.
• Output:
  • Retrained model download link.
  • Evaluation metrics (loss, accuracy).

3. Data Visualizations

• Page URL: /visualizations
• Features: Interactive plots highlighting dataset insights, such as:
  • Distribution of disease classes.
  • Feature correlations.
  • Trends in image quality or class balance.

4. Retrained Model Download

• Endpoint: /retrain/download-model

---

Flood Request Simulation

The application’s performance was evaluated under heavy load using Locust software. Results include:

• Average response time: X seconds
• Peak concurrent users: Y users

Detailed results are available:

• CSV File: Flood Simulation Results
• Visualizations:
  

---

Results and Evaluation

Prediction Example

Metric	Value
Predicted Class	Blight
Confidence Score	58.4%

Retraining Example

Metric	Value
Loss	0.56
Accuracy	83%

---

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests for improvements and new features.

---

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Let me know if further modifications are needed! 🚀