Automated Georeferencing and Extraction of Building Footprints from Remotely Sensed Imagery using Deep Learning

Aniruddha Khatua, Apratim Bhattacharya, Bharath Aithal

2024

Abstract

Extracting building footprints from remotely sensed photos is crucial in conducting analyses in domains such as land-use analysis, transportation planning and development, environmental studies, and others. Various methodologies and strategies have been suggested for extracting building footprints from satellite or UAV images, aiming to circumvent the arduous, time-consuming, less effective, and costly process of manually digitizing building footprints. These proposed methodologies and strategies have demonstrated their efficacy in detecting and extracting features. However, they do not adequately retain the geographical information during the output generation process. This paper presents a pipeline that can automatically extract geographical information from input photos and transfer it to the output image, thereby achieving automated georeferencing of the output image. The pipeline utilizes the YOLOV8 model, an advanced deep-learning-based architecture for object detection and segmentation. The detection and segmentation findings, combined with the acquired geographical information, are used to perform vectorization and generate vector images of the extracted building footprint. This suggested pipeline streamlines the process of obtaining building footprint data linked to geospatial information by automating the georeferencing and shapefile preparation phases, reducing the associated complications. This automation not only expedites the process but also improves the precision and uniformity of the output datasets.

Paper Citation

in Harvard Style

Khatua A., Bhattacharya A. and Aithal B. (2024). Automated Georeferencing and Extraction of Building Footprints from Remotely Sensed Imagery using Deep Learning. In Proceedings of the 10th International Conference on Geographical Information Systems Theory, Applications and Management - Volume 1: GISTAM; ISBN 978-989-758-694-1, SciTePress, pages 128-135. DOI: 10.5220/0012625300003696

in Bibtex Style

@conference{gistam24,
author={Aniruddha Khatua and Apratim Bhattacharya and Bharath Aithal},
title={Automated Georeferencing and Extraction of Building Footprints from Remotely Sensed Imagery using Deep Learning},
booktitle={Proceedings of the 10th International Conference on Geographical Information Systems Theory, Applications and Management - Volume 1: GISTAM},
year={2024},
pages={128-135},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012625300003696},
isbn={978-989-758-694-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 10th International Conference on Geographical Information Systems Theory, Applications and Management - Volume 1: GISTAM
TI - Automated Georeferencing and Extraction of Building Footprints from Remotely Sensed Imagery using Deep Learning
SN - 978-989-758-694-1
AU - Khatua A.
AU - Bhattacharya A.
AU - Aithal B.
PY - 2024
SP - 128
EP - 135
DO - 10.5220/0012625300003696
PB - SciTePress