Banca de QUALIFICAÇÃO: LÍDIA GABRIELLY DUTRA DE MENESES SANTOS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : LÍDIA GABRIELLY DUTRA DE MENESES SANTOS
DATE: 08/05/2026
TIME: 14:00
LOCAL: Sala de Reuniões do DCA
TITLE:

Multi-Variable Expanded Latent Space Autoencoder for Image Classification Applied to Visual Navigation of Unmanned Surface Vehicles

 

KEY WORDS:

Robotics Vision, Obstacle Detection, Image Analysis, Underwater and Surface Images, Multi-variable Autoencoder, Expanded Latent Space, Image Classification

 

PAGES: 37
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
SPECIALTY: Processamento Gráfico (Graphics)
SUMMARY:

We propose the use of the Multi-Variable Expanded Latent Space Autoencoder (MVELSA)
to classify aquatic imagery, encompassing both underwater and surface domains, to en-
able the autonomous navigation of Unmanned Surface Vehicles (USVs) in complex obstacle-
ridden scenarios. Our core hypothesis is that MVELSA can identify objects of interest
with efficacy and precision comparable or superior to traditional convolutional models
by leveraging an expanded latent representation that preserves critical morphological fea-
tures. To validate this, we employed two distinct datasets: the public AQUA20 bench-
mark, consisting of 20 subaquatic classes, and the SBSID (Sail-Boat Surface Images
Dataset), developed by our research group to feature floating obstacles. Experimental
results demonstrate that MVELSA achieves a macro-averaged F1-Score of 0.96, outper-
forming baseline models in handling highly imbalanced data. Integrated into the navi-
gation algorithms of a robotic sailboat, this system can facilitate autonomous movement
across oceans and lagoons with high reliability and minimal human intervention.

 

COMMITTEE MEMBERS:
Presidente - 1345674 - LUIZ MARCOS GARCIA GONCALVES
Externo ao Programa - 1170845 - BRUNO MARQUES FERREIRA DA SILVA - UFRNExterno à Instituição - EMERSON VILAR DE OLIVEIRA - UFRN