Logo Lanfrica
  • Home
  • Atlas
  • Insights
  • Docs
  • Sign in

© 2026 Lanfrica. All rights reserved. All copyrights of the resources shown on the Lanfrica website belong to the original copyright holders, unless explicitly stated otherwise.

Long-term visual localization in deep-sea underwater environment

Record type:

paperdataset
Creator:
BoiDunArnMar
Editor:
LabDYNInsUni
Publisher:
CCSD
Host:avatar
International audience With the advent of autonomous underwater vehicles comes the need to localize them precisely in their environment. The robot’s location is usually retrieved using acoustic positioning systems. However, in the context of autonomous vehicles, these systems may not be available or sufficiently accurate. A finer localization could be obtained using the robot’s visual observations. In this study, we benchmark state-of-the-art visual localization methods that were developed for terrestrial applications on the Eiffel Tower deepsea dataset. The latter embeds four visits of the same hydrothermal vent over five years. We show that these methods struggle to localize images collected in different years. We conduct an analysis to assess which factors may be responsible for this performance hit. Les récentes avancées liées aux véhicules sous-marins autonomes impliquent la nécessité de pouvoir localiser précisément ces derniers dans leur environnement. Cependant, la précision des systèmes de positionnement acoustique n'est pas suffisante. Une localisation plus fine pourrait alors être obtenue en utilisant les observations visuelles du robot. Dans cette étude, nous évaluons des méthodes de l'état de l'art de localisation visuelle développées en milieu terrestre sur le jeu de données sous-marin Eiffel Tower. Celui-ci contient des images issues de quatre visites de la même cheminée hydrothermale étendues sur cinq ans. Nous montrons que ces méthodes ont du mal à localiser des images issues d'années différentes. Nous menons ensuite une analyse pour évaluer les facteurs qui peuvent être responsables de cette baisse de performance.

Visit

hal.science

Tasks

computer vision

Tags

Visual localizationMarine roboticsLocalisation visuelleRobotique sous-marine[INFO.INFO-CV]Computer Science [cs]/Computer Vision and Pattern Recognition [cs.CV][INFO.INFO-RB]Computer Science [cs]/Robotics [cs.RO]

Licenses

info:eu-repo/semantics/OpenAccess

Similar

Long Term Variability of Extreme Sea Surface Temperature in the Gulf of Tonkin, ChinaSeasonal Rainfall Variability in Ethiopia and Its Long-Term Link to Global Sea Surface TemperaturesLong-Term Preservation for Access of Audio-Visual Archives at Botswana National Archives (BNARS)Endo-VMFuseNet: Deep Visual-Magnetic Sensor Fusion Approach for Uncalibrated, Unsynchronized and Asymmetric Endoscopic Capsule Robot Localization Datadeep-sea channel architecture knowledge graph deep-sea channel architecture knowledge graphShort-term Finance, Long-term Effects

Long Term Variability of Extreme Sea Surface Temperature in the Gulf of Tonkin, China

Till date, no study on trends in extreme sea surface temperature (SST) for different return periods

Seasonal Rainfall Variability in Ethiopia and Its Long-Term Link to Global Sea Surface Temperatures

Investigating the influence of sea surface temperatures (SSTs) on seasonal rainfall is a crucial fac

Long-Term Preservation for Access of Audio-Visual Archives at Botswana National Archives (BNARS)

This chapter discusses the long-term preservation and access to audio-visual (AV) archives at the Bo

Endo-VMFuseNet: Deep Visual-Magnetic Sensor Fusion Approach for Uncalibrated, Unsynchronized and Asymmetric Endoscopic Capsule Robot Localization Data

In the last decade, researchers and medical device companies have made major advances towards transf

deep-sea channel architecture knowledge graph deep-sea channel architecture knowledge graph

the data of deep-sea channel architecture knowledge graph for West African Nigeria offshore channel

Short-term Finance, Long-term Effects

We study the effect of short-term finance on firm growth and its aggregate implications in emerging