Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation

Extensive research studies have been conducted in recent years to exploit the complementarity among multisensor (or multimodal) remote sensing data for prominent applications such as land cover mapping. In order to make a step further with respect to previous studies, which investigate multitemporal...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Yawogan Jean Eudes Gbodjo, Olivier Montet, Dino Ienco, Raffaele Gaetano, Stephane Dupuy
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Convolutional neural networks (CNNs) land use and land cover (LULC) mapping multisensor multitemporal and multiscale remote sensing self-distillation sparsely annotated data Ocean engineering TC1501-1800 Geophysics. Cosmic physics QC801-809
Acceso en línea:	https://doaj.org/article/c122c482e24f42d599480cb0c63d64e6
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:c122c482e24f42d599480cb0c63d64e6
record_format	dspace
spelling	oai:doaj.org-article:c122c482e24f42d599480cb0c63d64e62021-11-20T00:00:23ZMultisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation2151-153510.1109/JSTARS.2021.3119191https://doaj.org/article/c122c482e24f42d599480cb0c63d64e62021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9566776/https://doaj.org/toc/2151-1535Extensive research studies have been conducted in recent years to exploit the complementarity among multisensor (or multimodal) remote sensing data for prominent applications such as land cover mapping. In order to make a step further with respect to previous studies, which investigate multitemporal SAR and optical data or multitemporal/multiscale optical combinations, here, we propose a deep learning framework that simultaneously integrates all these input sources, specifically multitemporal SAR/optical data and fine-scale optical information at their native temporal and spatial resolutions. Our proposal relies on a patch-based multibranch convolutional neural network (CNN) that exploits different per-source encoders to deal with the specificity of the input signals. In addition, we introduce a new self-distillation strategy to boost the per-source analyses and exploit the interplay among the different input sources. This new strategy leverages the final prediction of the multisource framework to guide the learning of the per-source CNN encoders supporting the network to learn from itself. Experiments are carried out on two real-world benchmarks, namely, the <italic>Reunion island</italic> (a French overseas department) and the <italic>Dordogne</italic> study site (a southwest department in France), where the annotated reference data were collected under operational constraints (sparsely annotated ground-truth data). Obtained results providing an overall classification accuracy of about 94% (respectively, 88%) on the <italic>Reunion island</italic> (respectively, the <italic>Dordogne</italic>) study site highlight the effectiveness of our framework based on CNNs and self-distillation to combine heterogeneous multisensor remote sensing data and confirm the benefit of multimodal analysis for downstream tasks such as land cover mapping.Yawogan Jean Eudes GbodjoOlivier MontetDino IencoRaffaele GaetanoStephane DupuyIEEEarticleConvolutional neural networks (CNNs)land use and land cover (LULC) mappingmultisensormultitemporal and multiscale remote sensingself-distillationsparsely annotated dataOcean engineeringTC1501-1800Geophysics. Cosmic physicsQC801-809ENIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Vol 14, Pp 11485-11499 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Convolutional neural networks (CNNs) land use and land cover (LULC) mapping multisensor multitemporal and multiscale remote sensing self-distillation sparsely annotated data Ocean engineering TC1501-1800 Geophysics. Cosmic physics QC801-809
spellingShingle	Convolutional neural networks (CNNs) land use and land cover (LULC) mapping multisensor multitemporal and multiscale remote sensing self-distillation sparsely annotated data Ocean engineering TC1501-1800 Geophysics. Cosmic physics QC801-809 Yawogan Jean Eudes Gbodjo Olivier Montet Dino Ienco Raffaele Gaetano Stephane Dupuy Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
description	Extensive research studies have been conducted in recent years to exploit the complementarity among multisensor (or multimodal) remote sensing data for prominent applications such as land cover mapping. In order to make a step further with respect to previous studies, which investigate multitemporal SAR and optical data or multitemporal/multiscale optical combinations, here, we propose a deep learning framework that simultaneously integrates all these input sources, specifically multitemporal SAR/optical data and fine-scale optical information at their native temporal and spatial resolutions. Our proposal relies on a patch-based multibranch convolutional neural network (CNN) that exploits different per-source encoders to deal with the specificity of the input signals. In addition, we introduce a new self-distillation strategy to boost the per-source analyses and exploit the interplay among the different input sources. This new strategy leverages the final prediction of the multisource framework to guide the learning of the per-source CNN encoders supporting the network to learn from itself. Experiments are carried out on two real-world benchmarks, namely, the <italic>Reunion island</italic> (a French overseas department) and the <italic>Dordogne</italic> study site (a southwest department in France), where the annotated reference data were collected under operational constraints (sparsely annotated ground-truth data). Obtained results providing an overall classification accuracy of about 94% (respectively, 88%) on the <italic>Reunion island</italic> (respectively, the <italic>Dordogne</italic>) study site highlight the effectiveness of our framework based on CNNs and self-distillation to combine heterogeneous multisensor remote sensing data and confirm the benefit of multimodal analysis for downstream tasks such as land cover mapping.
format	article
author	Yawogan Jean Eudes Gbodjo Olivier Montet Dino Ienco Raffaele Gaetano Stephane Dupuy
author_facet	Yawogan Jean Eudes Gbodjo Olivier Montet Dino Ienco Raffaele Gaetano Stephane Dupuy
author_sort	Yawogan Jean Eudes Gbodjo
title	Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
title_short	Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
title_full	Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
title_fullStr	Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
title_full_unstemmed	Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
title_sort	multisensor land cover classification with sparsely annotated data based on convolutional neural networks and self-distillation
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/c122c482e24f42d599480cb0c63d64e6
work_keys_str_mv	AT yawoganjeaneudesgbodjo multisensorlandcoverclassificationwithsparselyannotateddatabasedonconvolutionalneuralnetworksandselfdistillation AT oliviermontet multisensorlandcoverclassificationwithsparselyannotateddatabasedonconvolutionalneuralnetworksandselfdistillation AT dinoienco multisensorlandcoverclassificationwithsparselyannotateddatabasedonconvolutionalneuralnetworksandselfdistillation AT raffaelegaetano multisensorlandcoverclassificationwithsparselyannotateddatabasedonconvolutionalneuralnetworksandselfdistillation AT stephanedupuy multisensorlandcoverclassificationwithsparselyannotateddatabasedonconvolutionalneuralnetworksandselfdistillation
_version_	1718419861601779712

Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation

Ejemplares similares