Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite Images

Building type information indicates the functional properties of buildings and plays a crucial role in smart city development and urban socioeconomic activities. Existing methods for classifying building types often face challenges in accurately distinguishing buildings between types while maintaini...

Full description

Saved in:
Bibliographic Details
Main Authors: Shaofeng Zhang, Mengmeng Li, Wufan Zhao, Xiaoqin Wang, Qunyong Wu
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/10756709/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846128626094833664
author Shaofeng Zhang
Mengmeng Li
Wufan Zhao
Xiaoqin Wang
Qunyong Wu
author_facet Shaofeng Zhang
Mengmeng Li
Wufan Zhao
Xiaoqin Wang
Qunyong Wu
author_sort Shaofeng Zhang
collection DOAJ
description Building type information indicates the functional properties of buildings and plays a crucial role in smart city development and urban socioeconomic activities. Existing methods for classifying building types often face challenges in accurately distinguishing buildings between types while maintaining well-delineated boundaries, especially in complex urban environments. This study introduces a novel framework, i.e., CNN-Transformer cross-attention feature fusion network (CTCFNet), for building type classification from very high resolution remote sensing images. CTCFNet integrates convolutional neural networks (CNNs) and Transformers using an interactive cross-encoder fusion module that enhances semantic feature learning and improves classification accuracy in complex scenarios. We develop an adaptive collaboration optimization module that applies human visual attention mechanisms to enhance the feature representation of building types and boundaries simultaneously. To address the scarcity of datasets in building type classification, we create two new datasets, i.e., the urban building type (UBT) dataset and the town building type (TBT) dataset, for model evaluation. Extensive experiments on these datasets demonstrate that CTCFNet outperforms popular CNNs, Transformers, and dual-encoder methods in identifying building types across various regions, achieving the highest mean intersection over union of 78.20% and 77.11%, F1 scores of 86.83% and 88.22%, and overall accuracy of 95.07% and 95.73% on the UBT and TBT datasets, respectively. We conclude that CTCFNet effectively addresses the challenges of high interclass similarity and intraclass inconsistency in complex scenes, yielding results with well-delineated building boundaries and accurate building types.
format Article
id doaj-art-979926b8605b48b8b79e2e0649306ebc
institution Kabale University
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-979926b8605b48b8b79e2e0649306ebc2024-12-11T00:00:38ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-011897699410.1109/JSTARS.2024.350167810756709Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite ImagesShaofeng Zhang0https://orcid.org/0009-0001-0689-264XMengmeng Li1https://orcid.org/0000-0002-9083-0475Wufan Zhao2https://orcid.org/0000-0002-0265-3465Xiaoqin Wang3Qunyong Wu4Key Laboratory of Spatial Data Mining and Information Sharing of Ministry of Education, Academy of Digital China, Fuzhou University, Fuzhou, ChinaKey Laboratory of Spatial Data Mining and Information Sharing of Ministry of Education, Academy of Digital China, Fuzhou University, Fuzhou, ChinaUrban Governance and Design Thrust, Society Hub, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, ChinaKey Laboratory of Spatial Data Mining and Information Sharing of Ministry of Education, Academy of Digital China, Fuzhou University, Fuzhou, ChinaKey Laboratory of Spatial Data Mining and Information Sharing of Ministry of Education, Academy of Digital China, Fuzhou University, Fuzhou, ChinaBuilding type information indicates the functional properties of buildings and plays a crucial role in smart city development and urban socioeconomic activities. Existing methods for classifying building types often face challenges in accurately distinguishing buildings between types while maintaining well-delineated boundaries, especially in complex urban environments. This study introduces a novel framework, i.e., CNN-Transformer cross-attention feature fusion network (CTCFNet), for building type classification from very high resolution remote sensing images. CTCFNet integrates convolutional neural networks (CNNs) and Transformers using an interactive cross-encoder fusion module that enhances semantic feature learning and improves classification accuracy in complex scenarios. We develop an adaptive collaboration optimization module that applies human visual attention mechanisms to enhance the feature representation of building types and boundaries simultaneously. To address the scarcity of datasets in building type classification, we create two new datasets, i.e., the urban building type (UBT) dataset and the town building type (TBT) dataset, for model evaluation. Extensive experiments on these datasets demonstrate that CTCFNet outperforms popular CNNs, Transformers, and dual-encoder methods in identifying building types across various regions, achieving the highest mean intersection over union of 78.20% and 77.11%, F1 scores of 86.83% and 88.22%, and overall accuracy of 95.07% and 95.73% on the UBT and TBT datasets, respectively. We conclude that CTCFNet effectively addresses the challenges of high interclass similarity and intraclass inconsistency in complex scenes, yielding results with well-delineated building boundaries and accurate building types.https://ieeexplore.ieee.org/document/10756709/Building type classificationCNN-transformer networkscross-encoderfeature interactionvery high resolution remote sensing
spellingShingle Shaofeng Zhang
Mengmeng Li
Wufan Zhao
Xiaoqin Wang
Qunyong Wu
Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite Images
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Building type classification
CNN-transformer networks
cross-encoder
feature interaction
very high resolution remote sensing
title Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite Images
title_full Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite Images
title_fullStr Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite Images
title_full_unstemmed Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite Images
title_short Building Type Classification Using CNN-Transformer Cross-Encoder Adaptive Learning From Very High Resolution Satellite Images
title_sort building type classification using cnn transformer cross encoder adaptive learning from very high resolution satellite images
topic Building type classification
CNN-transformer networks
cross-encoder
feature interaction
very high resolution remote sensing
url https://ieeexplore.ieee.org/document/10756709/
work_keys_str_mv AT shaofengzhang buildingtypeclassificationusingcnntransformercrossencoderadaptivelearningfromveryhighresolutionsatelliteimages
AT mengmengli buildingtypeclassificationusingcnntransformercrossencoderadaptivelearningfromveryhighresolutionsatelliteimages
AT wufanzhao buildingtypeclassificationusingcnntransformercrossencoderadaptivelearningfromveryhighresolutionsatelliteimages
AT xiaoqinwang buildingtypeclassificationusingcnntransformercrossencoderadaptivelearningfromveryhighresolutionsatelliteimages
AT qunyongwu buildingtypeclassificationusingcnntransformercrossencoderadaptivelearningfromveryhighresolutionsatelliteimages