Exploring RAU-net for semantic segmentation of Philippines satellite images in identification of building density

Abstract

ABSTRACT This study explores RAU-Net convolutional network architecture on satellite images by classifying geospatial objects on satellite images such as roofs to help in determining the building density of a specific location. This study developed a satellite image dataset in the Philippines and trained the dataset in an RAU-Net Convolutional Neural Network using Tensorflow Keras API. This study proves that U-Networks with a few datasets and provides acceptable performance scores. Furthermore, a new way to calculate Building Density in terms of Building Coverage Ratio (BCR) using satellite images was also presented in this study. The model and the analysis were integrated into an application that determines the BCR of a specific location through a satellite map.