SatelliteImageClassification_CNN

This repository contains code and instructions for building a Convolutional Neural Network (CNN) to classify satellite images into Water, Green Area, Desert, and Cloudy with an accuracy of 95%. It includes the following steps:-

1. Importing the Dataset using Opendatasets
2. Splitting the data into training, validation and test sets
3. Defining the Convolutional Neural Network
4. Moving the model and data to GPU
5. Training the Model with fit and evaluate functions
6. Testing with individual images and calculating accuracy
7. Building a more efficient model using Transfer Learning (ResNet)

We will use the Satellite Image Classification Dataset comprising 5631 images categorized into four distinct classes: cloudy, desert, green area, and water.