<italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior Information

Window segmentation and vectorization remains a significant challenge, particularly in the absence of clean facade images. To extract complete window segments from building fa&#x00E7;ade images with occlusions, this article proposes an occlusion-aware window segmentation <italic>(OA-WinSeg...

Full description

Saved in:
Bibliographic Details
Main Authors: Manuela F. Ceron-Viveros, Wolfgang Maass, Jiaojiao Tian
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10923720/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849766863243837440
author Manuela F. Ceron-Viveros
Wolfgang Maass
Jiaojiao Tian
author_facet Manuela F. Ceron-Viveros
Wolfgang Maass
Jiaojiao Tian
author_sort Manuela F. Ceron-Viveros
collection DOAJ
description Window segmentation and vectorization remains a significant challenge, particularly in the absence of clean facade images. To extract complete window segments from building fa&#x00E7;ade images with occlusions, this article proposes an occlusion-aware window segmentation <italic>(OA-WinSeg)</italic> network with conditional adversarial training guided by prior structural information. This architecture combines the power of image segmentation and generative capabilities to handle occlusions. First, <italic>OA-WinSeg</italic> automatically detects occlusions and generates a rectangular boundary guidance from a coarse window segmentation, which incorporates structural information about the building layout into the process. Subsequently, the network refines the coarse segmentation and generates window segments in the missing regions by attending to contextual information of the nonoccluded parts of the fa&#x00E7;ade. Finally, our approach generates accurate vector representations, information needed for building modeling systems. Experimental results demonstrate the effectiveness of our model with simulated and occluded real-world datasets. In addition, we evaluate our model on various ablation studies to explore the contribution of the different modules. Finally, we have analyzed the potential applications of the proposed segmentation network and the completed window segments, including building fa&#x00E7;ade inpainting.
format Article
id doaj-art-06e23d31ac3e412e84760ebf888edab3
institution DOAJ
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-06e23d31ac3e412e84760ebf888edab32025-08-20T03:04:26ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01188490850310.1109/JSTARS.2025.355063210923720<italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior InformationManuela F. Ceron-Viveros0Wolfgang Maass1Jiaojiao Tian2https://orcid.org/0000-0002-8407-5098Department of Computer Science, Saarland University, Saarbr&#x00FC;cken, GermanyGerman Research Center for Artificial Intelligence, Kaiserslautern, GermanyRemote Sensing Technology Institute (IMF), German Aerospace Center (DLR), Wessling, GermanyWindow segmentation and vectorization remains a significant challenge, particularly in the absence of clean facade images. To extract complete window segments from building fa&#x00E7;ade images with occlusions, this article proposes an occlusion-aware window segmentation <italic>(OA-WinSeg)</italic> network with conditional adversarial training guided by prior structural information. This architecture combines the power of image segmentation and generative capabilities to handle occlusions. First, <italic>OA-WinSeg</italic> automatically detects occlusions and generates a rectangular boundary guidance from a coarse window segmentation, which incorporates structural information about the building layout into the process. Subsequently, the network refines the coarse segmentation and generates window segments in the missing regions by attending to contextual information of the nonoccluded parts of the fa&#x00E7;ade. Finally, our approach generates accurate vector representations, information needed for building modeling systems. Experimental results demonstrate the effectiveness of our model with simulated and occluded real-world datasets. In addition, we evaluate our model on various ablation studies to explore the contribution of the different modules. Finally, we have analyzed the potential applications of the proposed segmentation network and the completed window segments, including building fa&#x00E7;ade inpainting.https://ieeexplore.ieee.org/document/10923720/Attention modeldeep learningfaçade imagesinpaintingocclusionswindow segmentation
spellingShingle Manuela F. Ceron-Viveros
Wolfgang Maass
Jiaojiao Tian
<italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior Information
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Attention model
deep learning
façade images
inpainting
occlusions
window segmentation
title <italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior Information
title_full <italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior Information
title_fullStr <italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior Information
title_full_unstemmed <italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior Information
title_short <italic>OA-WinSeg:</italic> Occlusion-Aware Window Segmentation With Conditional Adversarial Training Guided by Structural Prior Information
title_sort italic oa winseg italic occlusion aware window segmentation with conditional adversarial training guided by structural prior information
topic Attention model
deep learning
façade images
inpainting
occlusions
window segmentation
url https://ieeexplore.ieee.org/document/10923720/
work_keys_str_mv AT manuelafceronviveros italicoawinsegitalicocclusionawarewindowsegmentationwithconditionaladversarialtrainingguidedbystructuralpriorinformation
AT wolfgangmaass italicoawinsegitalicocclusionawarewindowsegmentationwithconditionaladversarialtrainingguidedbystructuralpriorinformation
AT jiaojiaotian italicoawinsegitalicocclusionawarewindowsegmentationwithconditionaladversarialtrainingguidedbystructuralpriorinformation