The dataset comprises a set of images that includes different man-made objects considered critical in underwater operations. Anchors, buoy, chain, flutuation module, 3D marker, mine and pipeline were selected to represent some common objects that are presented in water scenarios. The data were acquired with a monocular camera in diverse conditions of the luminosity and perspectives of the objects. Moreover, the objects were captured in pure and turbid water and with white and coral reef background, to simulate real environments. Nowadays, the online available datasets are incomplete for this type scenarios and so, datasets as the UWObjects@CRAS are important to validate new perspectives and to answer to challenges, because to allow the development and validation of systems able to object identification, manipulation and obstacle avoidance, for example.

Data and Resources

Additional Info

Field Value
Author Alexandra Pereira Nunes
Maintainer Aníbal Matos
Last Updated May 2, 2019, 09:27 (UTC)
Created April 30, 2019, 10:30 (UTC)
Cite as NUNES, Alexandra Pereira, MATOS, Aníbal. UWObjects@CRAS [dataset].30 apr 2019. INESC TEC research data repository. DOI: https://doi.org/10.25747/60va-3c36
DOI https://doi.org/10.25747/60va-3c36
dc.Coverage.Spatial Test pool (indoor) in FEUP (Eletrical and Computer Engineering Department)
dc.Coverage.Temporal one day: January 7th
dc.Date 2019
dc.Format.Extent 1,92 GB
dc.Language EN
dc.Publisher INESC TEC
dc.Type Folders with monocular images of diverse (man-made) objects in underwater environment. The images were acquired in pure and turbid water. Moreover, the objects are captured in white and coral reef background (to simulate the real environment).
ddi.InstrumentName Monocular camera and critical objects (non-natural) - anchor, chain, mine, etc
ddi.InstrumentType Visual Sensor
ddi.Methodology Research on deep learning techniques for objects classification in underwater environemnt