Performance Comparison of Machine Learning Classifiers for Change Detection on Satellite Data

The proliferation of remotely sensed satellite data has significantly enhanced research in the field of Land Use and Land Cover (LULC) analysis. LULC analysis plays a crucial role in realizing the geographical features of the land in a region. Now a days, the rapid growth in development has triggered the need for precise predictions in the field of LULC change detection and thus has led to an increase in research work in the field. Hence, different bodies of research have surfaced analyzing different regions and machine learning models. However, the lack of practical implementation of machine learning models and their comparison in LULC change detection still prevails. In this study, data provided by the LANDSAT-8 satellite for 3 different years is leveraged for LULC classification and change detection within the administrative region encompassing Bhubaneswar and Cuttack. We employ a diverse array of machine learning techniques to compare and contrast the different algorithms by analyzing their classification accuracy rate on 10 different classes. The classification, sheds light on the interplay between human activities and natural landscapes, fostering informed decision-making for sustainable development, while the change detection provides vital information on the transforming and altering landscape of the region.