Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery

This paper proposes a new approach based on an unsupervised deep learning (DL) model for landslide detection. Recently, supervised DL models using convolutional neural networks (CNN) have been widely studied for landslide detection. Even though these models provide robust performance and reliable re...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Hejar Shahabi, Maryam Rahimzad, Sepideh Tavakkoli Piralilou, Omid Ghorbanzadeh, Saied Homayouni, Thomas Blaschke, Samsung Lim, Pedram Ghamisi
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	landslide mapping remote sensing unsupervised feature learning convolutional auto-encoder (CAE) mini-batch K-means Science Q
Acceso en línea:	https://doaj.org/article/c1812dc727b74ae59124462201e82446
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:c1812dc727b74ae59124462201e82446
record_format	dspace
spelling	oai:doaj.org-article:c1812dc727b74ae59124462201e824462021-11-25T18:55:31ZUnsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery10.3390/rs132246982072-4292https://doaj.org/article/c1812dc727b74ae59124462201e824462021-11-01T00:00:00Zhttps://www.mdpi.com/2072-4292/13/22/4698https://doaj.org/toc/2072-4292This paper proposes a new approach based on an unsupervised deep learning (DL) model for landslide detection. Recently, supervised DL models using convolutional neural networks (CNN) have been widely studied for landslide detection. Even though these models provide robust performance and reliable results, they depend highly on a large labeled dataset for their training step. As an alternative, in this paper, we developed an unsupervised learning model by employing a convolutional auto-encoder (CAE) to deal with the problem of limited labeled data for training. The CAE was used to learn and extract the abstract and high-level features without using training data. To assess the performance of the proposed approach, we used Sentinel-2 imagery and a digital elevation model (DEM) to map landslides in three different case studies in India, China, and Taiwan. Using minimum noise fraction (MNF) transformation, we reduced the multispectral dimension to three features containing more than 80% of scene information. Next, these features were stacked with slope data and NDVI as inputs to the CAE model. The Huber reconstruction loss was used to evaluate the inputs. We achieved reconstruction losses ranging from 0.10 to 0.147 for the MNF features, slope, and NDVI stack for all three study areas. The mini-batch K-means clustering method was used to cluster the features into two to five classes. To evaluate the impact of deep features on landslide detection, we first clustered a stack of MNF features, slope, and NDVI, then the same ones plus with the deep features. For all cases, clustering based on deep features provided the highest precision, recall, F1-score, and mean intersection over the union in landslide detection.Hejar ShahabiMaryam RahimzadSepideh Tavakkoli PiralilouOmid GhorbanzadehSaied HomayouniThomas BlaschkeSamsung LimPedram GhamisiMDPI AGarticlelandslide mappingremote sensingunsupervised feature learningconvolutional auto-encoder (CAE)mini-batch K-meansScienceQENRemote Sensing, Vol 13, Iss 4698, p 4698 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	landslide mapping remote sensing unsupervised feature learning convolutional auto-encoder (CAE) mini-batch K-means Science Q
spellingShingle	landslide mapping remote sensing unsupervised feature learning convolutional auto-encoder (CAE) mini-batch K-means Science Q Hejar Shahabi Maryam Rahimzad Sepideh Tavakkoli Piralilou Omid Ghorbanzadeh Saied Homayouni Thomas Blaschke Samsung Lim Pedram Ghamisi Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery
description	This paper proposes a new approach based on an unsupervised deep learning (DL) model for landslide detection. Recently, supervised DL models using convolutional neural networks (CNN) have been widely studied for landslide detection. Even though these models provide robust performance and reliable results, they depend highly on a large labeled dataset for their training step. As an alternative, in this paper, we developed an unsupervised learning model by employing a convolutional auto-encoder (CAE) to deal with the problem of limited labeled data for training. The CAE was used to learn and extract the abstract and high-level features without using training data. To assess the performance of the proposed approach, we used Sentinel-2 imagery and a digital elevation model (DEM) to map landslides in three different case studies in India, China, and Taiwan. Using minimum noise fraction (MNF) transformation, we reduced the multispectral dimension to three features containing more than 80% of scene information. Next, these features were stacked with slope data and NDVI as inputs to the CAE model. The Huber reconstruction loss was used to evaluate the inputs. We achieved reconstruction losses ranging from 0.10 to 0.147 for the MNF features, slope, and NDVI stack for all three study areas. The mini-batch K-means clustering method was used to cluster the features into two to five classes. To evaluate the impact of deep features on landslide detection, we first clustered a stack of MNF features, slope, and NDVI, then the same ones plus with the deep features. For all cases, clustering based on deep features provided the highest precision, recall, F1-score, and mean intersection over the union in landslide detection.
format	article
author	Hejar Shahabi Maryam Rahimzad Sepideh Tavakkoli Piralilou Omid Ghorbanzadeh Saied Homayouni Thomas Blaschke Samsung Lim Pedram Ghamisi
author_facet	Hejar Shahabi Maryam Rahimzad Sepideh Tavakkoli Piralilou Omid Ghorbanzadeh Saied Homayouni Thomas Blaschke Samsung Lim Pedram Ghamisi
author_sort	Hejar Shahabi
title	Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery
title_short	Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery
title_full	Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery
title_fullStr	Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery
title_full_unstemmed	Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery
title_sort	unsupervised deep learning for landslide detection from multispectral sentinel-2 imagery
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/c1812dc727b74ae59124462201e82446
work_keys_str_mv	AT hejarshahabi unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery AT maryamrahimzad unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery AT sepidehtavakkolipiralilou unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery AT omidghorbanzadeh unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery AT saiedhomayouni unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery AT thomasblaschke unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery AT samsunglim unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery AT pedramghamisi unsuperviseddeeplearningforlandslidedetectionfrommultispectralsentinel2imagery
_version_	1718410550549938176

Unsupervised Deep Learning for Landslide Detection from Multispectral Sentinel-2 Imagery

Ejemplares similares