A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPING
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-wor...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2023-04-01
|
| Series: | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| Online Access: | https://isprs-archives.copernicus.org/articles/XLVIII-M-1-2023/217/2023/isprs-archives-XLVIII-M-1-2023-217-2023.pdf |
| _version_ | 1797842596587896832 |
|---|---|
| author | M. X. Ortega Adarme M. X. Ortega Adarme P. J. Soto Vega G. A. O. P. Costa R. Q. Feitosa C. Heipke |
| author_facet | M. X. Ortega Adarme M. X. Ortega Adarme P. J. Soto Vega G. A. O. P. Costa R. Q. Feitosa C. Heipke |
| author_sort | M. X. Ortega Adarme |
| collection | DOAJ |
| description | 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. |
| first_indexed | 2024-04-09T16:50:33Z |
| format | Article |
| id | doaj.art-c67115d9f8e04b3c8331be9d86d72ca6 |
| institution | Directory Open Access Journal |
| issn | 1682-1750 2194-9034 |
| language | English |
| last_indexed | 2024-04-09T16:50:33Z |
| publishDate | 2023-04-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| spelling | doaj.art-c67115d9f8e04b3c8331be9d86d72ca62023-04-21T15:04:13ZengCopernicus PublicationsThe International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences1682-17502194-90342023-04-01XLVIII-M-1-202321722310.5194/isprs-archives-XLVIII-M-1-2023-217-2023A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPINGM. X. Ortega Adarme0M. X. Ortega Adarme1P. J. Soto Vega2G. A. O. P. Costa3R. Q. Feitosa4C. Heipke5Institute of Photogrammetry and GeoInformation, Leibniz Universität Hannover, GermanyPontifical Catholic University of Rio de Janeiro (PUC-Rio), BrazilLaTIM, INSERM, UMR 1101, University Brest, Brest, FranceState University of Rio de Janeiro (UERJ), Rio de Janeiro, BrazilPontifical Catholic University of Rio de Janeiro (PUC-Rio), BrazilInstitute of Photogrammetry and GeoInformation, Leibniz Universität Hannover, GermanyDeep 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.https://isprs-archives.copernicus.org/articles/XLVIII-M-1-2023/217/2023/isprs-archives-XLVIII-M-1-2023-217-2023.pdf |
| spellingShingle | M. X. Ortega Adarme M. X. Ortega Adarme P. J. Soto Vega G. A. O. P. Costa R. Q. Feitosa C. Heipke A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPING The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| title | A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPING |
| title_full | A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPING |
| title_fullStr | A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPING |
| title_full_unstemmed | A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPING |
| title_short | A DEBIASING VARIATIONAL AUTOENCODER FOR DEFORESTATION MAPPING |
| title_sort | debiasing variational autoencoder for deforestation mapping |
| url | https://isprs-archives.copernicus.org/articles/XLVIII-M-1-2023/217/2023/isprs-archives-XLVIII-M-1-2023-217-2023.pdf |
| work_keys_str_mv | AT mxortegaadarme adebiasingvariationalautoencoderfordeforestationmapping AT mxortegaadarme adebiasingvariationalautoencoderfordeforestationmapping AT pjsotovega adebiasingvariationalautoencoderfordeforestationmapping AT gaopcosta adebiasingvariationalautoencoderfordeforestationmapping AT rqfeitosa adebiasingvariationalautoencoderfordeforestationmapping AT cheipke adebiasingvariationalautoencoderfordeforestationmapping AT mxortegaadarme debiasingvariationalautoencoderfordeforestationmapping AT mxortegaadarme debiasingvariationalautoencoderfordeforestationmapping AT pjsotovega debiasingvariationalautoencoderfordeforestationmapping AT gaopcosta debiasingvariationalautoencoderfordeforestationmapping AT rqfeitosa debiasingvariationalautoencoderfordeforestationmapping AT cheipke debiasingvariationalautoencoderfordeforestationmapping |