Satellite_Image_Classification_By_CNN

For satellite image classification, I use a convolutional neural network (cNN) architecture. I collected the data from a publicly available data source. First of all, I preprocess the data and clean it, then I convert the data source into a CSV file. Then I split the data set between 80% for training and 20% for testing. After splitting, I run the deep learning model using TensorFlow and Keras, especially in CNN architecture. I use the convolutional 2D architecture and max pulling for this model. For the activation function, I use Relu in the deep neural network model and the softmax activation function in the output layer. I ran the DNN model only for five epochs, and my output accuracy was around 85%. I find different things in the data and make some visualizations. I added all the visualizations to the Readme.MD folder. I found the loss and accuracy visualization for training and validation split, and then I found some other things from the data, like wanting to make a confusion matrix for this data set. In the advanced prediction model, I want to create a system where if we input data from a URL, we will find the output of the image using our classification model.

Findings of our Model :

Dataset values	Dataset time series	Dataset faceted distributions

Dataset distributions	categorical distributions	Accuracy Graph

Loss Graph	Confussion Matrix	Epoches Accuracy

CNN Architecture of this model :