Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement Handling

This study presents an enhanced Faster R-CNN framework that incorporates elliptical bounding boxes to significantly improve building detection in off-nadir imagery, effectively reducing severe geometric distortions caused by oblique sensor angles. Off-nadir imagery enhances architectural detail capt...

Full description

Saved in:
Bibliographic Details
Main Authors: Sejung Jung, Ahram Song, Kirim Lee, Won Hee Lee
Format: Article
Language:English
Published: MDPI AG 2025-04-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/17/7/1247
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849730373514166272
author Sejung Jung
Ahram Song
Kirim Lee
Won Hee Lee
author_facet Sejung Jung
Ahram Song
Kirim Lee
Won Hee Lee
author_sort Sejung Jung
collection DOAJ
description This study presents an enhanced Faster R-CNN framework that incorporates elliptical bounding boxes to significantly improve building detection in off-nadir imagery, effectively reducing severe geometric distortions caused by oblique sensor angles. Off-nadir imagery enhances architectural detail capture and reduces occlusions, but conventional bounding boxes, such as axis-aligned and rotated bounding boxes, often fail to localize buildings distorted by extreme perspectives. We propose a hybrid method integrating elliptical bounding boxes for curved structures and rotated bounding boxes for tilted buildings, achieving more precise shape approximation. In addition, our model incorporates a squeeze-and-excitation mechanism to refine feature representation, suppress background noise, and enhance object boundary alignment, leading to superior detection accuracy. Experimental results on the BONAI dataset demonstrate that our approach achieves a detection rate of 91.96%, significantly outperforming axis-aligned bounding boxes (65.75%) and rotated bounding boxes (87.13%) in detecting irregular and distorted buildings. By providing a highly robust and adaptable detection strategy, our approach establishes a new standard for accurate and shape-aware building recognition in off-nadir imagery, significantly improving the detection of distorted, rotated, and irregular structures.
format Article
id doaj-art-2e1bda51e0154f9ca8297789ea4e9622
institution DOAJ
issn 2072-4292
language English
publishDate 2025-04-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj-art-2e1bda51e0154f9ca8297789ea4e96222025-08-20T03:08:54ZengMDPI AGRemote Sensing2072-42922025-04-01177124710.3390/rs17071247Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement HandlingSejung Jung0Ahram Song1Kirim Lee2Won Hee Lee3Department of Convergence and Fusion System Engineering, Kyungpook National University, Sangju 37224, Republic of KoreaDepartment of Location-Based Information System, Kyungpook National University, Sangju 37224, Republic of KoreaResearch Institute of Artificial Intelligent Diagnosis Technology for Multi-Scale Organic and Inorganic Structure, Kyungpook National University, Sangju 37224, Republic of KoreaDepartment of Location-Based Information System, Kyungpook National University, Sangju 37224, Republic of KoreaThis study presents an enhanced Faster R-CNN framework that incorporates elliptical bounding boxes to significantly improve building detection in off-nadir imagery, effectively reducing severe geometric distortions caused by oblique sensor angles. Off-nadir imagery enhances architectural detail capture and reduces occlusions, but conventional bounding boxes, such as axis-aligned and rotated bounding boxes, often fail to localize buildings distorted by extreme perspectives. We propose a hybrid method integrating elliptical bounding boxes for curved structures and rotated bounding boxes for tilted buildings, achieving more precise shape approximation. In addition, our model incorporates a squeeze-and-excitation mechanism to refine feature representation, suppress background noise, and enhance object boundary alignment, leading to superior detection accuracy. Experimental results on the BONAI dataset demonstrate that our approach achieves a detection rate of 91.96%, significantly outperforming axis-aligned bounding boxes (65.75%) and rotated bounding boxes (87.13%) in detecting irregular and distorted buildings. By providing a highly robust and adaptable detection strategy, our approach establishes a new standard for accurate and shape-aware building recognition in off-nadir imagery, significantly improving the detection of distorted, rotated, and irregular structures.https://www.mdpi.com/2072-4292/17/7/1247off-nadir imagerybuilding detectionelliptical bounding boxesrotated bounding boxesaxis-aligned bounding boxesgeometric distortion
spellingShingle Sejung Jung
Ahram Song
Kirim Lee
Won Hee Lee
Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement Handling
Remote Sensing
off-nadir imagery
building detection
elliptical bounding boxes
rotated bounding boxes
axis-aligned bounding boxes
geometric distortion
title Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement Handling
title_full Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement Handling
title_fullStr Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement Handling
title_full_unstemmed Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement Handling
title_short Advanced Building Detection with Faster R-CNN Using Elliptical Bounding Boxes for Displacement Handling
title_sort advanced building detection with faster r cnn using elliptical bounding boxes for displacement handling
topic off-nadir imagery
building detection
elliptical bounding boxes
rotated bounding boxes
axis-aligned bounding boxes
geometric distortion
url https://www.mdpi.com/2072-4292/17/7/1247
work_keys_str_mv AT sejungjung advancedbuildingdetectionwithfasterrcnnusingellipticalboundingboxesfordisplacementhandling
AT ahramsong advancedbuildingdetectionwithfasterrcnnusingellipticalboundingboxesfordisplacementhandling
AT kirimlee advancedbuildingdetectionwithfasterrcnnusingellipticalboundingboxesfordisplacementhandling
AT wonheelee advancedbuildingdetectionwithfasterrcnnusingellipticalboundingboxesfordisplacementhandling