Application of Geospatial Artificial Intelligence in Building Footprint Extraction from Aerial Imagery Using Deep Learning

Abstract

This study employs Geospatial Artificial Intelligence (GeoAI) applications to analyze and detect geographic features by extracting building footprints from aerial imagery using Deep Learning techniques. The approach aims to achieve a high level of accuracy and efficiency in mapping and geographic feature extraction. The study seeks to enhance the capabilities of Geographic Information Systems (GIS) software in processing large volumes of geospatial data and monitoring spatial changes that occur over specific time periods. To achieve these objectives, a deep learning-based building footprint extraction approach was applied to the Al-Nuaimah area in Irbid Governorate as a case study. Two sets of aerial images were analyzed: one acquired in 2007 and another in 2024. The study integrated Geospatial Artificial Intelligence (GeoAI), GIS applications, and Remote Sensing (RS) techniques to identify and quantify changes in the built environment over the study period. The findings revealed a noticeable increase in the number of buildings between 2007 and 2024, demonstrating significant urban growth within the study area. The results also confirmed the effectiveness of GeoAI applications in analyzing and extracting geographic features from aerial imagery with a high degree of accuracy and efficiency. The study recommends the wider adoption of deep learning approaches in geographical research, given their remarkable success in automated feature extraction and mapping tasks, which remain challenging for traditional machine learning methods.