Klasifikasi Bangunan secara Otomatis Menggunakan Pembelajaran Mendalam dari Gambar Street-View

Abstract

Urban population density mapping or urban utility planning requires a classification map based on individual buildings that are considered much more informative. The goal of this research is to determine how to extract the fine-grained boundaries of individual buildings from a street-view dataset. This paper proposes a general framework for classifying individual building functionality using a deep learning approach. The proposed method is based on a Convolutional Neural Network (CNN) that classifies facade structures from street view images, such as Street-View images. From the experiments conducted, the CNN classifier with the ResNet architecture was able to classify the Street-View data group with an accuracy value of 86.79%. We construct a dataset to train and evaluate the CNN classifier. Furthermore, the method is applied to generate a building classification map at the urban area scale.