SATELLITE IMAGE CLASSIFICATION USING MACHINE LEARNING

Kilari Veera Swamy; Polugumati Raasi; Savaraju Mounika; P. Veeshalakshi

Abstract

Satellite imagery is used in for various applications such as disaster management, defence, surveys, and environmental understanding. Earlier, manual processing was applied to do analysis. These methods are not accurate to detect the target and classify the object. Hence, in this work machine learning is explored. Now a days deep learning has gain prominence to automate target detection and classification. Convolutional Neural Networks (CNN) are playing vital role in deep learning techniques. In this work, simple deep learning model is built. It consists of 2 CNN layers. After each convolution maximum pooling operation is used. One flattening layer is used. After flattening, 2 fully connected layers are used. At the end SoftMax is used to classify the satellite images. Various weights are computed during training. Accordingly, deep CNN model is built. The same built model is used for testing. Experimental results indicate that proposed structure is able to classify images accurately