ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest Mapping
Although deep learning techniques are known to achieve outstanding classification accuracies, remote sensing datasets often present limited labeled data and class imbalances, two challenges to attaining high levels of accuracy. In recent years, the GAN architecture has achieved great success as a da...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10141632/ |
| _version_ | 1797330427931787264 |
|---|---|
| author | Alvaro G. Dieste Francisco Arguello Dora B. Heras |
| author_facet | Alvaro G. Dieste Francisco Arguello Dora B. Heras |
| author_sort | Alvaro G. Dieste |
| collection | DOAJ |
| description | Although deep learning techniques are known to achieve outstanding classification accuracies, remote sensing datasets often present limited labeled data and class imbalances, two challenges to attaining high levels of accuracy. In recent years, the GAN architecture has achieved great success as a data augmentation method, driving research toward further enhancements. This work presents ResBaGAN, a GAN-based method for the classification of remote sensing images, designed to overcome the challenges of data scarcity and class imbalances by constructing an advanced data augmentation framework. This framework builds upon a GAN architecture enhanced with an autoencoder initialization and class balancing properties, a superpixel-based sample extraction procedure with traditional augmentation techniques, and an improved residual network as classifier. Experiments were conducted on large, very high-resolution multispectral images of riparian forests in Galicia, Spain, with limited training data and strong class imbalances, comparing ResBaGAN to other machine learning methods such as simpler GANs. ResBaGAN achieved higher overall classification accuracies, particularly improving the accuracy of minority classes with F1-score enhancements reaching up to 22%. |
| first_indexed | 2024-03-08T07:19:41Z |
| format | Article |
| id | doaj.art-bd73225e2e874c0b9c44b90f28ad9cc5 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-03-08T07:19:41Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-bd73225e2e874c0b9c44b90f28ad9cc52024-02-03T00:00:14ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352023-01-01166428644710.1109/JSTARS.2023.328189210141632ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest MappingAlvaro G. Dieste0https://orcid.org/0000-0002-6429-7283Francisco Arguello1https://orcid.org/0000-0001-9279-5426Dora B. Heras2https://orcid.org/0000-0002-5304-1426Centro Singular de Investigación en Tecnoloxías Intelixentes, Universidade de Santiago de Compostela, Santiago de Compostela, SpainDepartamento de Electrónica e Computación, Universidade de Santiago de Compostela, Santiago de Compostela, SpainCentro Singular de Investigación en Tecnoloxías Intelixentes, Universidade de Santiago de Compostela, Santiago de Compostela, SpainAlthough deep learning techniques are known to achieve outstanding classification accuracies, remote sensing datasets often present limited labeled data and class imbalances, two challenges to attaining high levels of accuracy. In recent years, the GAN architecture has achieved great success as a data augmentation method, driving research toward further enhancements. This work presents ResBaGAN, a GAN-based method for the classification of remote sensing images, designed to overcome the challenges of data scarcity and class imbalances by constructing an advanced data augmentation framework. This framework builds upon a GAN architecture enhanced with an autoencoder initialization and class balancing properties, a superpixel-based sample extraction procedure with traditional augmentation techniques, and an improved residual network as classifier. Experiments were conducted on large, very high-resolution multispectral images of riparian forests in Galicia, Spain, with limited training data and strong class imbalances, comparing ResBaGAN to other machine learning methods such as simpler GANs. ResBaGAN achieved higher overall classification accuracies, particularly improving the accuracy of minority classes with F1-score enhancements reaching up to 22%.https://ieeexplore.ieee.org/document/10141632/BAGANclassificationdata augmentationmultispectralresidual networksuperpixels |
| spellingShingle | Alvaro G. Dieste Francisco Arguello Dora B. Heras ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest Mapping IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing BAGAN classification data augmentation multispectral residual network superpixels |
| title | ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest Mapping |
| title_full | ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest Mapping |
| title_fullStr | ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest Mapping |
| title_full_unstemmed | ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest Mapping |
| title_short | ResBaGAN: A Residual Balancing GAN with Data Augmentation for Forest Mapping |
| title_sort | resbagan a residual balancing gan with data augmentation for forest mapping |
| topic | BAGAN classification data augmentation multispectral residual network superpixels |
| url | https://ieeexplore.ieee.org/document/10141632/ |
| work_keys_str_mv | AT alvarogdieste resbaganaresidualbalancingganwithdataaugmentationforforestmapping AT franciscoarguello resbaganaresidualbalancingganwithdataaugmentationforforestmapping AT dorabheras resbaganaresidualbalancingganwithdataaugmentationforforestmapping |