An Implementation of Satellite Image Classification and Analysis using Machine Learning with ISRO LISS IV

Abstract

For several years, detecting objects in satellite imagery has been a difficult task. Higher accuracy in the identification of different artifacts from very high-resolution satellite images has been achieved thanks to the creation of successful machine learning algorithms and advancements in hardware systems. Satellite imaging has been successfully used for weather forecasts and spatial and geological purposes over the last few decades. For these types of applications, low- resolution satellite images are adequate. In our project, high-resolution images will be used where the prediction is done by processing their images and producing respective valid data for the same. The high-resolution satellite images are being provided by RRSC- C ISRO, Nagpur. For autonomous systems to interact with their environment intelligently, they must be given the ability to adapt and learn incrementally and deliberately. This project focuses on extracting and identification of objects from the high-resolution satellite images which will be provided as an input from the satellite to extract the features of the image which are then converted into machine valid data. This valid data is then fed to the neural network model. The neural network model analyzes each image and classifies it with the feature label. This will generate the output screen which displays the extracted feature from the original image which is given as input to the system.