Details
| Originalsprache | Englisch |
|---|---|
| Seiten (von - bis) | 217-223 |
| Seitenumfang | 7 |
| Fachzeitschrift | International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives |
| Jahrgang | XLVIII-M-1-2023 |
| Publikationsstatus | Veröffentlicht - 21 Apr. 2023 |
| Veranstaltung | 39th International Symposium on Remote Sensing of Environment, ISRSE 2023 - Antalya, Türkei Dauer: 24 Apr. 2023 → 28 Apr. 2023 |
Abstract
Deep Learning (DL) algorithms provide numerous benefits in different applications, and they usually yield successful results in scenarios with enough labeled training data and similar class proportions. However, the labeling procedure is a cost and time-consuming task. Furthermore, numerous real-world classification problems present a high level of class imbalance, as the number of samples from the classes of interest differ significantly. In various cases, such conditions tend to promote the creation of biased systems, which negatively impact their performance. Designing unbiased systems has been an active research topic, and recently some DL-based techniques have demonstrated encouraging results in that regard. In this work, we introduce an extension of the Debiasing Variational Autoencoder (DB-VAE) for semantic segmentation. The approach is based on an end-to-end DL scheme and employs the learned latent variables to adjust the individual sampling probabilities of data points during the training process. For that purpose, we adapted the original DB-VAE architecture for dense labeling in the context of deforestation mapping. Experiments were carried out on a region of the Brazilian Amazon, using Sentinel-2 data and the deforestation map from the PRODES project. The reported results show that the proposed DB-VAE approach is able to learn and identify under-represented samples, and select them more frequently in the training batches, consequently delivering superior classification metrics.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Information systems
- Sozialwissenschaften (insg.)
- Geografie, Planung und Entwicklung
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
in: International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives, Jahrgang XLVIII-M-1-2023, 21.04.2023, S. 217-223.
Publikation: Beitrag in Fachzeitschrift › Konferenzaufsatz in Fachzeitschrift › Forschung › Peer-Review
}
TY - JOUR
T1 - A debiasing variational autoencoder for deforestation mapping
AU - Adarme, M. X.Ortega
AU - Vega, P. J.Soto
AU - Costa, G. A.O.P.
AU - Feitosa, R. Q.
AU - Heipke, C.
N1 - Funding Information: The authors would like to thank the German Academic Exchange Service (DAAD), the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), the Coordenac¸ão de Aperfeic¸oamento de Pessoal de Nível Superior (CAPES), and the Fundac¸ão de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) for the financial support.
PY - 2023/4/21
Y1 - 2023/4/21
N2 - Deep Learning (DL) algorithms provide numerous benefits in different applications, and they usually yield successful results in scenarios with enough labeled training data and similar class proportions. However, the labeling procedure is a cost and time-consuming task. Furthermore, numerous real-world classification problems present a high level of class imbalance, as the number of samples from the classes of interest differ significantly. In various cases, such conditions tend to promote the creation of biased systems, which negatively impact their performance. Designing unbiased systems has been an active research topic, and recently some DL-based techniques have demonstrated encouraging results in that regard. In this work, we introduce an extension of the Debiasing Variational Autoencoder (DB-VAE) for semantic segmentation. The approach is based on an end-to-end DL scheme and employs the learned latent variables to adjust the individual sampling probabilities of data points during the training process. For that purpose, we adapted the original DB-VAE architecture for dense labeling in the context of deforestation mapping. Experiments were carried out on a region of the Brazilian Amazon, using Sentinel-2 data and the deforestation map from the PRODES project. The reported results show that the proposed DB-VAE approach is able to learn and identify under-represented samples, and select them more frequently in the training batches, consequently delivering superior classification metrics.
AB - Deep Learning (DL) algorithms provide numerous benefits in different applications, and they usually yield successful results in scenarios with enough labeled training data and similar class proportions. However, the labeling procedure is a cost and time-consuming task. Furthermore, numerous real-world classification problems present a high level of class imbalance, as the number of samples from the classes of interest differ significantly. In various cases, such conditions tend to promote the creation of biased systems, which negatively impact their performance. Designing unbiased systems has been an active research topic, and recently some DL-based techniques have demonstrated encouraging results in that regard. In this work, we introduce an extension of the Debiasing Variational Autoencoder (DB-VAE) for semantic segmentation. The approach is based on an end-to-end DL scheme and employs the learned latent variables to adjust the individual sampling probabilities of data points during the training process. For that purpose, we adapted the original DB-VAE architecture for dense labeling in the context of deforestation mapping. Experiments were carried out on a region of the Brazilian Amazon, using Sentinel-2 data and the deforestation map from the PRODES project. The reported results show that the proposed DB-VAE approach is able to learn and identify under-represented samples, and select them more frequently in the training batches, consequently delivering superior classification metrics.
KW - Debiasing Variational Autoencoder
KW - Deep Learning
KW - Deforestation Detection
KW - Semantic Segmentation
UR - http://www.scopus.com/inward/record.url?scp=85156261731&partnerID=8YFLogxK
U2 - 10.5194/isprs-archives-XLVIII-M-1-2023-217-2023
DO - 10.5194/isprs-archives-XLVIII-M-1-2023-217-2023
M3 - Conference article
AN - SCOPUS:85156261731
VL - XLVIII-M-1-2023
SP - 217
EP - 223
JO - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
JF - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
SN - 1682-1750
T2 - 39th International Symposium on Remote Sensing of Environment, ISRSE 2023
Y2 - 24 April 2023 through 28 April 2023
ER -