[DATASET] Remote sensing imagery of threatened Vereda wetlands in the Brazilian Midwest: a segmentation-oriented model as a backbone for domain-specific applications

Fuente: Zenodo
Guardado en:
Detalles Bibliográficos
Autores principales: Machicao, Jeaneth, Barbosa, Alexandre Augusto, O. Salles, Leandro, Toledo, Peter, Corrêa, Pedro, Oliveira, Luiz, Collevatti, Rosane, de Souza, Eduardo Barroso, Ometto, Jean
Formato: Recurso digital
Lenguaje:inglés
Publicado: Zenodo 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866901804126568448
author Machicao, Jeaneth
Barbosa, Alexandre Augusto
O. Salles, Leandro
Toledo, Peter
Corrêa, Pedro
Oliveira, Luiz
Collevatti, Rosane
de Souza, Eduardo Barroso
Ometto, Jean
author_facet Machicao, Jeaneth
Barbosa, Alexandre Augusto
O. Salles, Leandro
Toledo, Peter
Corrêa, Pedro
Oliveira, Luiz
Collevatti, Rosane
de Souza, Eduardo Barroso
Ometto, Jean
contents <p>This repository provides the annotated dataset and pretrained U-Net model for segmenting Vereda wetlands (VED.SGF) in the Brazilian Cerrado. The <code>data/model/</code> directory contains the final architecture (<code>model.keras</code>) and weights (<code>model.weights.h5</code>). The <code>data/train/</code> folder includes the RGB mosaic (<code>box_train_4m_mosaic.tif</code>), annotation masks (<code>box_train_4m_labels.tif</code>), and training/validation files (<code>train.h5</code>, <code>validation.h5</code>). The <code>data/predict/</code> folder provides an example input tile and its corresponding prediction. Shapefiles defining training/test regions (<code>shapefiles/LearningBox</code>) and the study area (<code>shapefiles/StudyArea</code>) are also included. </p> <p>Downloading this dataset complements the source code available on our GitHub (<a href="https://github.com/wetlands-vision/wetlands-segmentation-unet">https://github.com/wetlands-vision/wetlands-segmentation-unet</a>), enabling full reproducibility of training, evaluation, and inference workflows.</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_17803750
institution Zenodo
language eng
publishDate 2025
publisher Zenodo
record_format zenodo
spellingShingle [DATASET] Remote sensing imagery of threatened Vereda wetlands in the Brazilian Midwest: a segmentation-oriented model as a backbone for domain-specific applications
Machicao, Jeaneth
Barbosa, Alexandre Augusto
O. Salles, Leandro
Toledo, Peter
Corrêa, Pedro
Oliveira, Luiz
Collevatti, Rosane
de Souza, Eduardo Barroso
Ometto, Jean
semantic segmentation
U-net
wetland mapping
cerrado vereda ecosystems
remote sensing
deep learning
<p>This repository provides the annotated dataset and pretrained U-Net model for segmenting Vereda wetlands (VED.SGF) in the Brazilian Cerrado. The <code>data/model/</code> directory contains the final architecture (<code>model.keras</code>) and weights (<code>model.weights.h5</code>). The <code>data/train/</code> folder includes the RGB mosaic (<code>box_train_4m_mosaic.tif</code>), annotation masks (<code>box_train_4m_labels.tif</code>), and training/validation files (<code>train.h5</code>, <code>validation.h5</code>). The <code>data/predict/</code> folder provides an example input tile and its corresponding prediction. Shapefiles defining training/test regions (<code>shapefiles/LearningBox</code>) and the study area (<code>shapefiles/StudyArea</code>) are also included. </p> <p>Downloading this dataset complements the source code available on our GitHub (<a href="https://github.com/wetlands-vision/wetlands-segmentation-unet">https://github.com/wetlands-vision/wetlands-segmentation-unet</a>), enabling full reproducibility of training, evaluation, and inference workflows.</p>
title [DATASET] Remote sensing imagery of threatened Vereda wetlands in the Brazilian Midwest: a segmentation-oriented model as a backbone for domain-specific applications
topic semantic segmentation
U-net
wetland mapping
cerrado vereda ecosystems
remote sensing
deep learning
url https://doi.org/10.5281/zenodo.17803750