Background/Objectives: Growth in urbanization correlates with the continuous expansion of towns and cities. Unchecked, this growth poses a serious threat to the environment and the boundaries of urban areas. An increase in metropolitan space calls for rapid and constant supervision to track sustainable development. For the purposes of urban planning and analysis, remote sensing and GIS are extremely important. The unstructured data that comes with a building’s machine learning and image processing aids in automating the architectural feature extraction process. Although this manual extraction is significant for urban and energy tasks, it is also very costly and labour-intensive. The purpose of this paper is to use autonomous building extraction algorithms to generate a dataset of building footprints. Methods: The methodology encompasses data preprocessing, segmentation of built-up areas via thresholding, and computing the results. This methodology was carried out on a dataset consisting of CartoSAT-2 images of Gandhinagar City. Findings: The efficacy of the methodology was evaluated with three separate testing datasets, each having 207 building footprints extracted. The proposed method achieved extraction accuracies of 83.10%, 92.38%, and 89.52% in the three test regions. The complete building footprint detection F1-score was 0.9267, which demonstrates a balanced precision and recall. Novelty: This study demonstrates a novel way of building footprint extraction that is automated through the use of a thresholding algorithm, thereby reducing manual work. In contrast with conventional GIS methods, it applies a combination of machine learning and image processing, which enhances accuracy and scalability. Keywords: Automated Extraction of Building Footprints, Cartosat-2, Land Use, Geographic Information System, Urban Development
Patel et al. (Wed,) studied this question.