Image Classification of Land Use Land Cover of Bengaluru City Using Convolutional Neural Network

Abstract

Developing countries like India is witnessing an increasing economic growth, rapid population in addition to industrialization leading to an increased rate of land use and cover. In order to better utilize the land and natural resource is essential to classify and analyse the land use and cover. Machine Learning and Deep Learning techniques are considered to be one of the effective and efficient ways for analysing and classifying the land use & cover. Here, in this paper, methodology for land use & cover classification – analysis of rural and urban regions of Bengaluru is been proposed. The proposed system’s main objective is to monitor the land cover changes of Bengaluru district including its rural and urban region for classifying the land cover into its exact classes. Classification algorithms such as SVM (Support Vector Machine), RF (Random Forest), KNN (K – Nearest Neighbor) and DT (Decision Tree) are used in the preprocessing of images and model created is tested using CNN. The Landsat datasets from usgs earth explorer is used. Performance evaluation of these algorithms are done based on their accuracy rates and efficiency. The proposed system shows that CNN classifies the land cover classes efficiently because of its highest accuracy and efficiency rates when compared with other algorithms.