Convolutional neural networks for hydrothermal vents substratum classification: An introspective study 

PJS Vega, P Papadakis, M Matabos, L Van Audenhaege, A Ramiere, J Sarrazin, GAOP da Costa.

Ecological Informatics. 

This work investigates the application of state-of-the-art network architectures, originally employed in the classification of non-seabed images, for the task of hydrothermal vent substrata image classification. In assessing their potential, we conduct a study on the generalization, complementarity, and human interpretability aspects of those architectures.

Weakly supervised adversarial domain adaptation for deforestation detection in tropical forests

PJS Vega, GAOP Costa, MXO Adarme, JDB Castro, RQ Feitosa.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 

This work proposes a weakly-supervised, adversarial domain adaptation method for a change detection task based on the Domain Adversarial Neural Network (DANN) strategy. In short, DANN aligns features in a domain-agnostic space through adversarial learning, which often implies the deterioration of feature discriminability as a side effect of the adversarial alignment, which does not take into consideration class labels of the target domain samples.

Domain-adversarial neural networks for deforestation detection in tropical forests

Pedro J Soto, Gilson A Costa, Raul Q Feitosa, Mabel X Adarme, José D Burmudez, Javier N Turnes.

IEEE Geoscience and Remote Sensing Letters. 

In this work, we propose a DL-based representation matching approach for DA in the context of change detection tasks. We further evaluate the approach in a deforestation mapping application, characterized by a high-class imbalance between the deforestation and no-deforestation classes. The main motivation raised when observing the performance of such representation-matching techniques is negatively impacted when class occurrences in the target domain, for which no labeled data are available during training, are highly imbalanced.

Deforestation detection with fully convolutional networks in the Amazon forest from Landsat-8 and Sentinel-2 images

DL Torres, JN Turnes, PJS Vega, RQ Feitosa, DE Silva, JM Junior, C Almeida.

Remote Sensing, MDPI. 

This paper comprehensively explores state-of-the-art fully convolutional networks such as U-Net, ResU-Net, SegNet, FC-DenseNet, and two DeepLabv3+ variants for monitoring deforestation in the Brazilian Amazon. The networks’ performance is evaluated experimentally in terms of Precision, Recall, F1-score, and computational load using satellite images with different spatial and spectral resolutions: Landsat-8 and Sentinel-2. 

An unsupervised domain adaptation approach for change detection and its application to deforestation mapping in tropical biomes

PJS Vega, GAOP da Costa, RQ Feitosa, MXO Adarme, CA de Almeida, C Heipke, FRottensteiner.

ISPRS Journal of Photogrammetry and Remote Sensing. 

This paper proposes a new domain adaptation approach for change detection based on the CycleGAN model. To the best of our knowledge, this is the first work that addresses unsupervised DA for change detection. The results were obtained by evaluating over three different sites in the Amazon and Brazilian Cerrado biomes.

Atrous cGAN for SAR to Optical Image Translation.

JN Turnes, JDB Castro,  DL Torres, PJS Vega, RQ Feitosa, PN Happ.

IEEE Geoscience and Remote Sensing Letters.

The paper reports experiments carried out to assess the performance of atrous-cGAN for the synthesis of Landsat-8 images from Sentinel-1A data based on three public datasets. The experimental analysis indicated that the \emph{atrous-cGAN} consistently outperformed the classical  pix2pix counterpart in terms of visual quality, similarity with the true optical image, and as a feature learning tool for semantic segmentation.

A Deviasing Variational Autoencoder For Deforestation Mapping.

MX Ortega Adarme, PJ Soto Vega, GAOP Costa, RQ Feitosa, C Hepike.

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

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.

Deforestation Detection With Weak Supervised Convolutional Neural Networks In Tropical Biomes.

PJ Soto Vega, GAOP Costa, MX Ortega Adarme, JD Bermudez, RQ Feitosa.

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

This work aims at investigating a noisy-label-based weak supervised method in the context of a deforestation mapping application, characterized by a high class imbalance between the classes of interest, i.e., deforestation and no-deforestation. The study sites correspond to different regions in the Amazon and Brazilian Cerrado biomes. 

Adversarial Discriminative Domain Adaptation For Deforestation Detection.

JN Turnes, PJ Soto, GAOP Costa, D Wittich, RQ Feitosa, F Rottensteiner.

ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences 

In this work, we introduce a domain adaptation approach based on representation matching for the deforestation detection task. The approach follows the Adversarial Discriminative Domain Adaptation (ADDA) framework, and we introduce a margin-based regularization constraint in the learning process that promotes a better convergence of the model parameters during training. The approach is evaluated using three different domains, which represent sites in different forest biomes. 

Domain Adaptation with CycleGAN for Change Detection in the Amazon Forest.

PJ Soto, G Costa, RQ Feitosa, PN Happ, MX Ortega, J Noa, CA Almeida, C Heipke.

The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences .

In this work, we exploit the CycleGAN approach for domain adaptation in a particular change detection application, namely, deforestation detection in the Amazon forest. Experimental results indicate that the proposed approach is capable of alleviating the effects associated with domain shift in the context of the target application.

Evaluation of Semantic Segmentation Methods For Deforestation Detection in the Amazon.

RB Andrade, G Costa, GLA Mota, MX Ortega, RQ Feitosa, PJ Soto, C Heipke.

The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences .

In this work, we implement and evaluate a deforestation detection approach which is based on a Fully Convolutional, Deep Learning (DL) model: the DeepLabv3+. We compare the results obtained with the devised approach to those obtained with previously proposed DL-based methods (Early Fusion and Siamese Convolutional Network) using Landsat OLI-8 images acquired at different dates, covering a region of the Amazon forest 

A Comparative Analysis Of Unsupervised And Semi-Supervised Representation Learning For Remote Sensing Image Categorization.

PJ Soto, JD Bermudez, PN Happ, RQ Feitosa.

ISPRS Annals of Photogrammetry, Remote Sensing & Spatial Information Sciences.

This work aims at investigating unsupervised and semi-supervised representation learning methods based on generative adversarial networks for remote sensing scene classification. The work introduces a novel approach,  which consists in a semi-supervised extension of a prior unsupervised method, known as MARTA-GAN. The proposed method delivered in our analysis the best overall accuracy under scarce labeled samples, both in terms of absolute value and in terms of variability across multiple runs.

Single Sample Face Recognition From Video via Stacked Supervised Auto-encoder.

PJS Vega, RQ Feitosa, VHA Quirita, PN Happ.

CONFERENCE ON GRAPHICS, PATTERNS AND IMAGES, 29. (SIBGRAPI) 

This work proposes and evaluates strategies based on Stacked Supervised Auto-Encoders (SSAE) for face representation in video surveillance applications. The study focuses on the identification task with a single sample per person (SSPP) in the gallery. Variations in terms of pose, facial expression, illumination and occlusion are approached in two ways. First, the SSAE extracts features from face images, which are robust to such variations. Second, we propose methods to exploit the multiple samples per persons probes (MSPPP) that can be extracted from video sequences. 

A Fuzzy Inference System for Multispectyral Image Classification.

PJS Vega, VAA Quirita, PM Achancaray, R Tanscheit, M Vellasco

2016 IEEE ANDESCON

This work presents an approach for multispectral image classification that makes use of a Fuzzy Inference System (FIS). An IKONOS satellite sensor image of a neighborhood in Rio de Janeiro, Brazil has been used. The ground truth used in this work comprises six classes: trees, scrub, buildings, roads, water and shadows. 

Face Likelihood Functions For Visual Tracking In Intelligent Spaces.

 Frank Sanabria-Macías, Enrique Marañón-Reyes, Pedro Soto-Vega, Marta Marrón-Romera, Javier Macias-Guarasa, Daniel Pizarro-Perez.

IECON2013-39th Annual Conference of the IEEE Industrial Electronics Society.

In this work we explore some Viola and Jones based likelihood functions presented in literature, and propose new strategies. We also extend the evaluation of the likelihood functions in position, scale and pose. One of our proposed functions shows better characteristics to be used in intelligent spaces in three dimensional face tracking applications.