Details
| Originalsprache | Englisch |
|---|---|
| Aufsatznummer | 044504 |
| Fachzeitschrift | Journal of applied remote sensing |
| Jahrgang | 15 |
| Ausgabenummer | 4 |
| Publikationsstatus | Veröffentlicht - 15 Okt. 2021 |
Abstract
The Brazilian Savanna, also known as Cerrado, is considered a global hotspot for biodiversity conservation. The detailed mapping of vegetation types, called physiognomies, is still a challenge due to their high spectral similarity and spatial variability. There are three major ecosystem groups (forest, savanna, and grassland), which can be hierarchically subdivided into 25 detailed physiognomies, according to a well-known classification system. We used an adapted U-net architecture to process a WorldView-2 image with 2-m spatial resolution to hierarchically classify the physiognomies of a Cerrado protected area based on deep learning techniques. Several spectral channels were tested as input datasets to classify the three major ecosystem groups (first level of classification). The dataset composed of RGB bands plus 2-band enhanced vegetation index (EVI2) achieved the best performance and was used to perform the hierarchical classification. In the first level of classification, the overall accuracy was 92.8%. On the other hand, for the savanna and grassland detailed physiognomies (second level of classification), 86.1% and 85.0% were reached, respectively. As the first work that intended to classify Cerrado physiognomies in this level of detail using deep learning, our accuracy rates outperformed others that applied traditional machine learning algorithms for this task.
ASJC Scopus Sachgebiete
- Erdkunde und Planetologie (insg.)
- Allgemeine Erdkunde und Planetologie
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
in: Journal of applied remote sensing, Jahrgang 15, Nr. 4, 044504, 15.10.2021.
Publikation: Beitrag in Fachzeitschrift › Artikel › Forschung › Peer-Review
}
TY - JOUR
T1 - Hierarchical mapping of Brazilian Savanna (Cerrado) physiognomies based on deep learning
AU - Neves, Alana K.
AU - Körting, Thales S.
AU - Fonseca, Leila M.G.
AU - Soares, Anderson R.
AU - Girolamo-Neto, Cesare D.
AU - Heipke, Christian
N1 - Funding Information: The authors thank the DigitalGlobe Foundation for supplying the image for this work and cooperating with INPE. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES) – Finance Code 001 and the Brazilian National Research Council (CNPq) (Grant Nos. 140372/2017-2 and 303360/2019-4). Part of this study was carried out while the first author was working at IPI (Institut für Photogrammetrie und GeoInformation), Leibniz Universität Hannover, Germany, with a scholarship from the Deutscher Akademischer Austauschdienst (DAAD). The authors declare no conflict of interest.
PY - 2021/10/15
Y1 - 2021/10/15
N2 - The Brazilian Savanna, also known as Cerrado, is considered a global hotspot for biodiversity conservation. The detailed mapping of vegetation types, called physiognomies, is still a challenge due to their high spectral similarity and spatial variability. There are three major ecosystem groups (forest, savanna, and grassland), which can be hierarchically subdivided into 25 detailed physiognomies, according to a well-known classification system. We used an adapted U-net architecture to process a WorldView-2 image with 2-m spatial resolution to hierarchically classify the physiognomies of a Cerrado protected area based on deep learning techniques. Several spectral channels were tested as input datasets to classify the three major ecosystem groups (first level of classification). The dataset composed of RGB bands plus 2-band enhanced vegetation index (EVI2) achieved the best performance and was used to perform the hierarchical classification. In the first level of classification, the overall accuracy was 92.8%. On the other hand, for the savanna and grassland detailed physiognomies (second level of classification), 86.1% and 85.0% were reached, respectively. As the first work that intended to classify Cerrado physiognomies in this level of detail using deep learning, our accuracy rates outperformed others that applied traditional machine learning algorithms for this task.
AB - The Brazilian Savanna, also known as Cerrado, is considered a global hotspot for biodiversity conservation. The detailed mapping of vegetation types, called physiognomies, is still a challenge due to their high spectral similarity and spatial variability. There are three major ecosystem groups (forest, savanna, and grassland), which can be hierarchically subdivided into 25 detailed physiognomies, according to a well-known classification system. We used an adapted U-net architecture to process a WorldView-2 image with 2-m spatial resolution to hierarchically classify the physiognomies of a Cerrado protected area based on deep learning techniques. Several spectral channels were tested as input datasets to classify the three major ecosystem groups (first level of classification). The dataset composed of RGB bands plus 2-band enhanced vegetation index (EVI2) achieved the best performance and was used to perform the hierarchical classification. In the first level of classification, the overall accuracy was 92.8%. On the other hand, for the savanna and grassland detailed physiognomies (second level of classification), 86.1% and 85.0% were reached, respectively. As the first work that intended to classify Cerrado physiognomies in this level of detail using deep learning, our accuracy rates outperformed others that applied traditional machine learning algorithms for this task.
KW - Cerrado
KW - physiognomy
KW - protected area
KW - Savanna
KW - semantic segmentation
KW - spectral channels
UR - http://www.scopus.com/inward/record.url?scp=85122695610&partnerID=8YFLogxK
U2 - 10.1117/1.JRS.15.044504
DO - 10.1117/1.JRS.15.044504
M3 - Article
AN - SCOPUS:85122695610
VL - 15
JO - Journal of applied remote sensing
JF - Journal of applied remote sensing
SN - 1931-3195
IS - 4
M1 - 044504
ER -