satelliteImageClassification

Python and MATLAB codes for satellite image classification using a convolutional neural network.

The dataset used was gotten from

https://www.kaggle.com/datasets/mahmoudreda55/satellite-image-classification

Image Classification Repository

This repository contains Python and MATLAB code for classifying satellite images using a convolutional neural network (CNN). The dataset used is available at Kaggle: Satellite Image Classification.

Project Overview

The project demonstrates how to build and train a CNN to classify satellite images. It includes:

• Data Preprocessing: Loading and preparing the dataset for training.
• Model Architecture: Defining the CNN structure.
• Training: Training the model on the dataset.
• Evaluation: Assessing the model's performance on test data.

Repository Structure

• LICENSE: License information.
• README.md: Project overview and instructions.
• satellite.ipynb: Jupyter Notebook with Python code for training and evaluating the CNN.
• satellite_images.mlx: MATLAB Live Script for training and evaluating the CNN.

Getting Started

To run the code:

1. Clone the repository:

   git clone https://github.com/al-chris/satelliteImageClassification.git

2. Navigate to the project directory:

   cd satelliteImageClassification

3. Install the required Python packages:

   Ensure you have Python installed, then install the necessary packages:

   pip install -r requirements.txt

   Note: The requirements.txt file should list all required packages. If it's missing, refer to the imports in satellite.ipynb and install the packages manually.

4. Download the dataset:

   Download the dataset from Kaggle and extract it into the project directory.

5. Run the Jupyter Notebook:

   jupyter notebook satellite.ipynb

   Follow the instructions in the notebook to train and evaluate the model.

6. For MATLAB users:

   Open satellite_images.mlx in MATLAB and run the script to train and evaluate the model.

Model Architecture

The CNN architecture includes:

• Convolutional layers
• Batch normalization
• ReLU activation functions
• Max pooling layers
• Fully connected (dense) layers

Visualization

  _________________
 |      Input      |
 |      Data       |
 |_________________|
         |
  _________________
 |     Conv2D      |
 |_________________|
 |   BatchNorm2D   |
 |_________________|
 |      ReLU       |
 |_________________|
 |  MaxPooling2D   |
 |_________________|
         |
  _________________
 |     Conv2D      |
 |_________________|
 |   BatchNorm2D   |
 |_________________|
 |      ReLU       |
 |_________________|
 |  MaxPooling2D   |
 |_________________|
         |
  _________________
 |     Conv2D      |
 |_________________|
 |  BatchNorm2D    |
 |_________________|
 |      ReLU       |
 |_________________|
 |     Flatten     |
 |_________________|
         |
  _________________
 |      Dense      |
 |_________________|

This structure is designed to effectively capture spatial hierarchies in the input images.

Results

After training, the model achieves an accuracy of approximately 98% on the test set.

Contributing

Contributions are welcome. Feel free to open issues or submit pull requests.

License

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

Acknowledgments

• Dataset: Kaggle: Satellite Image Classification

For any questions or suggestions, please contact Christopher Aliu.