Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion

Differential interferometric synthetic aperture radar (DInSAR), coherence, phase, and displacement are derived from processing SAR images to monitor geological phenomena and urban change. Previously, Sentinel-1 SAR data combined with Sentinel-2 optical imagery has improved classification accuracy in...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Zainoolabadien Karim, Terence L. van Zyl
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	machine learning deep learning transfer learning artificial intelligence remote sensing earth observation Science Q
Acceso en línea:	https://doaj.org/article/375fcd9d882c4ff7a8e83c3f7d15e68f
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:375fcd9d882c4ff7a8e83c3f7d15e68f
record_format	dspace
spelling	oai:doaj.org-article:375fcd9d882c4ff7a8e83c3f7d15e68f2021-11-11T18:55:24ZDeep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion10.3390/rs132143942072-4292https://doaj.org/article/375fcd9d882c4ff7a8e83c3f7d15e68f2021-10-01T00:00:00Zhttps://www.mdpi.com/2072-4292/13/21/4394https://doaj.org/toc/2072-4292Differential interferometric synthetic aperture radar (DInSAR), coherence, phase, and displacement are derived from processing SAR images to monitor geological phenomena and urban change. Previously, Sentinel-1 SAR data combined with Sentinel-2 optical imagery has improved classification accuracy in various domains. However, the fusing of Sentinel-1 DInSAR processed imagery with Sentinel-2 optical imagery has not been thoroughly investigated. Thus, we explored this fusion in urban change detection by creating a verified balanced binary classification dataset comprising 1440 blobs. Machine learning models using feature descriptors and non-deep learning classifiers, including a two-layer convolutional neural network (ConvNet2), were used as baselines. Transfer learning by feature extraction (TLFE) using various pre-trained models, deep learning from random initialization, and transfer learning by fine-tuning (TLFT) were all evaluated. We introduce a feature space ensemble family (FeatSpaceEnsNet), an average ensemble family (AvgEnsNet), and a hybrid ensemble family (HybridEnsNet) of TLFE neural networks. The FeatSpaceEnsNets combine TLFE features directly in the feature space using logistic regression. AvgEnsNets combine TLFEs at the decision level by aggregation. HybridEnsNets are a combination of FeatSpaceEnsNets and AvgEnsNets. Several FeatSpaceEnsNets, AvgEnsNets, and HybridEnsNets, comprising a heterogeneous mixture of different depth and architecture models, are defined and evaluated. We show that, in general, TLFE outperforms both TLFT and classic deep learning for the small dataset used and that larger ensembles of TLFE models do not always improve accuracy. The best performing ensemble is an AvgEnsNet (84.862%) comprised of a ResNet50, ResNeXt50, and EfficientNet B4. This was matched by a similarly composed FeatSpaceEnsNet with an F1 score of 0.001 and variance of 0.266 less. The best performing HybridEnsNet had an accuracy of 84.775%. All of the ensembles evaluated outperform the best performing single model, ResNet50 with TLFE (83.751%), except for AvgEnsNet 3, AvgEnsNet 6, and FeatSpaceEnsNet 5. Five of the seven similarly composed FeatSpaceEnsNets outperform the corresponding AvgEnsNet.Zainoolabadien KarimTerence L. van ZylMDPI AGarticlemachine learningdeep learningtransfer learningartificial intelligenceremote sensingearth observationScienceQENRemote Sensing, Vol 13, Iss 4394, p 4394 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	machine learning deep learning transfer learning artificial intelligence remote sensing earth observation Science Q
spellingShingle	machine learning deep learning transfer learning artificial intelligence remote sensing earth observation Science Q Zainoolabadien Karim Terence L. van Zyl Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion
description	Differential interferometric synthetic aperture radar (DInSAR), coherence, phase, and displacement are derived from processing SAR images to monitor geological phenomena and urban change. Previously, Sentinel-1 SAR data combined with Sentinel-2 optical imagery has improved classification accuracy in various domains. However, the fusing of Sentinel-1 DInSAR processed imagery with Sentinel-2 optical imagery has not been thoroughly investigated. Thus, we explored this fusion in urban change detection by creating a verified balanced binary classification dataset comprising 1440 blobs. Machine learning models using feature descriptors and non-deep learning classifiers, including a two-layer convolutional neural network (ConvNet2), were used as baselines. Transfer learning by feature extraction (TLFE) using various pre-trained models, deep learning from random initialization, and transfer learning by fine-tuning (TLFT) were all evaluated. We introduce a feature space ensemble family (FeatSpaceEnsNet), an average ensemble family (AvgEnsNet), and a hybrid ensemble family (HybridEnsNet) of TLFE neural networks. The FeatSpaceEnsNets combine TLFE features directly in the feature space using logistic regression. AvgEnsNets combine TLFEs at the decision level by aggregation. HybridEnsNets are a combination of FeatSpaceEnsNets and AvgEnsNets. Several FeatSpaceEnsNets, AvgEnsNets, and HybridEnsNets, comprising a heterogeneous mixture of different depth and architecture models, are defined and evaluated. We show that, in general, TLFE outperforms both TLFT and classic deep learning for the small dataset used and that larger ensembles of TLFE models do not always improve accuracy. The best performing ensemble is an AvgEnsNet (84.862%) comprised of a ResNet50, ResNeXt50, and EfficientNet B4. This was matched by a similarly composed FeatSpaceEnsNet with an F1 score of 0.001 and variance of 0.266 less. The best performing HybridEnsNet had an accuracy of 84.775%. All of the ensembles evaluated outperform the best performing single model, ResNet50 with TLFE (83.751%), except for AvgEnsNet 3, AvgEnsNet 6, and FeatSpaceEnsNet 5. Five of the seven similarly composed FeatSpaceEnsNets outperform the corresponding AvgEnsNet.
format	article
author	Zainoolabadien Karim Terence L. van Zyl
author_facet	Zainoolabadien Karim Terence L. van Zyl
author_sort	Zainoolabadien Karim
title	Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion
title_short	Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion
title_full	Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion
title_fullStr	Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion
title_full_unstemmed	Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion
title_sort	deep/transfer learning with feature space ensemble networks (featspaceensnets) and average ensemble networks (avgensnets) for change detection using dinsar sentinel-1 and optical sentinel-2 satellite data fusion
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/375fcd9d882c4ff7a8e83c3f7d15e68f
work_keys_str_mv	AT zainoolabadienkarim deeptransferlearningwithfeaturespaceensemblenetworksfeatspaceensnetsandaverageensemblenetworksavgensnetsforchangedetectionusingdinsarsentinel1andopticalsentinel2satellitedatafusion AT terencelvanzyl deeptransferlearningwithfeaturespaceensemblenetworksfeatspaceensnetsandaverageensemblenetworksavgensnetsforchangedetectionusingdinsarsentinel1andopticalsentinel2satellitedatafusion
_version_	1718431636274544640

Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion

Ejemplares similares