Identification of alkaline fens using convolutional neural networks and multispectral satellite imagery
The alkaline fen is a particularly valuable type of wetland with unique characteristics.Due to anthropogenic risk factors and the sensitive nature of the fens, protection is highlyprioritized with identification and mapping of current locations being important parts ofthis process. To accomplish thi...
Main Author: | |
---|---|
Format: | Others |
Language: | English |
Published: |
Luleå tekniska universitet, Datavetenskap
2021
|
Subjects: | |
Online Access: | http://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-87426 |
id |
ndltd-UPSALLA1-oai-DiVA.org-ltu-87426 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-UPSALLA1-oai-DiVA.org-ltu-874262021-10-14T05:24:15ZIdentification of alkaline fens using convolutional neural networks and multispectral satellite imageryengJernberg, JohnLuleå tekniska universitet, Datavetenskap2021Machine learningDeep learningRemote sensingData scienceSatelliteSentinelComputer SciencesDatavetenskap (datalogi)Computer Vision and Robotics (Autonomous Systems)Datorseende och robotik (autonoma system)Remote SensingFjärranalysteknikThe alkaline fen is a particularly valuable type of wetland with unique characteristics.Due to anthropogenic risk factors and the sensitive nature of the fens, protection is highlyprioritized with identification and mapping of current locations being important parts ofthis process. To accomplish this in a cost effective manner for large areas, remote sensingmethods using satellite images might be very effective. Following the rapid developmentin computer vision, deep learning using convolutional neural networks (CNN) is thecurrent state of the art for satellite image classification. Accordingly, this study evaluatesthe combination of different CNN architectures and multispectral Sentinel 2 satelliteimages for identification of alkaline fens using semantic segmentation. The implementedmodels are different variations of the proven U-net network design. In addition, a RandomForest classifier was trained for baseline comparison. The best result was produced bya spatial attention U-net with a IoU-score of 0.31 for the alkaline fen class and a meanIoU-score of 0.61. These findings suggest that identification of alkaline fens is possiblewith the current method even with a small dataset. However, an optimal solution tothis task may require deeper research. The results also further establish deep learningto be the superior choice over traditional machine learning algorithms for satellite imageclassification. Student thesisinfo:eu-repo/semantics/bachelorThesistexthttp://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-87426application/pdfinfo:eu-repo/semantics/openAccess |
collection |
NDLTD |
language |
English |
format |
Others
|
sources |
NDLTD |
topic |
Machine learning Deep learning Remote sensing Data science Satellite Sentinel Computer Sciences Datavetenskap (datalogi) Computer Vision and Robotics (Autonomous Systems) Datorseende och robotik (autonoma system) Remote Sensing Fjärranalysteknik |
spellingShingle |
Machine learning Deep learning Remote sensing Data science Satellite Sentinel Computer Sciences Datavetenskap (datalogi) Computer Vision and Robotics (Autonomous Systems) Datorseende och robotik (autonoma system) Remote Sensing Fjärranalysteknik Jernberg, John Identification of alkaline fens using convolutional neural networks and multispectral satellite imagery |
description |
The alkaline fen is a particularly valuable type of wetland with unique characteristics.Due to anthropogenic risk factors and the sensitive nature of the fens, protection is highlyprioritized with identification and mapping of current locations being important parts ofthis process. To accomplish this in a cost effective manner for large areas, remote sensingmethods using satellite images might be very effective. Following the rapid developmentin computer vision, deep learning using convolutional neural networks (CNN) is thecurrent state of the art for satellite image classification. Accordingly, this study evaluatesthe combination of different CNN architectures and multispectral Sentinel 2 satelliteimages for identification of alkaline fens using semantic segmentation. The implementedmodels are different variations of the proven U-net network design. In addition, a RandomForest classifier was trained for baseline comparison. The best result was produced bya spatial attention U-net with a IoU-score of 0.31 for the alkaline fen class and a meanIoU-score of 0.61. These findings suggest that identification of alkaline fens is possiblewith the current method even with a small dataset. However, an optimal solution tothis task may require deeper research. The results also further establish deep learningto be the superior choice over traditional machine learning algorithms for satellite imageclassification. |
author |
Jernberg, John |
author_facet |
Jernberg, John |
author_sort |
Jernberg, John |
title |
Identification of alkaline fens using convolutional neural networks and multispectral satellite imagery |
title_short |
Identification of alkaline fens using convolutional neural networks and multispectral satellite imagery |
title_full |
Identification of alkaline fens using convolutional neural networks and multispectral satellite imagery |
title_fullStr |
Identification of alkaline fens using convolutional neural networks and multispectral satellite imagery |
title_full_unstemmed |
Identification of alkaline fens using convolutional neural networks and multispectral satellite imagery |
title_sort |
identification of alkaline fens using convolutional neural networks and multispectral satellite imagery |
publisher |
Luleå tekniska universitet, Datavetenskap |
publishDate |
2021 |
url |
http://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-87426 |
work_keys_str_mv |
AT jernbergjohn identificationofalkalinefensusingconvolutionalneuralnetworksandmultispectralsatelliteimagery |
_version_ |
1719489762884059136 |