This paper presents a deep learning approach to detect underwater bubbles, coming from the seafloor, at different locations. The You Only Look Once (YOLO) framework has been used, modified and adapted for bubble detection. Training has been performed on images from the location of interest in Vulcano, Italy and on other images found on internet. The model was then trained on a Tesla T4 GPU by Google Colaboratory. The trained model has produced encouraging results that can be used for real-time bubble detection

A Deep Learning Approach for Underwater Bubble Detection / Bhattarai P.; Krupinski S.; Unnithan V.; Maurelli F.; Secciani N.; Franchi M.; Zacchini L.; Ridolfi A.. - ELETTRONICO. - 2021-September:(2021), pp. 1-5. ((Intervento presentato al convegno OCEANS 2021: San Diego – Porto tenutosi a San Diego, USA nel 20-23 settembre 2021 [10.23919/OCEANS44145.2021.9706107].

A Deep Learning Approach for Underwater Bubble Detection

Secciani N.;Zacchini L.
;
Ridolfi A.
2021

Abstract

This paper presents a deep learning approach to detect underwater bubbles, coming from the seafloor, at different locations. The You Only Look Once (YOLO) framework has been used, modified and adapted for bubble detection. Training has been performed on images from the location of interest in Vulcano, Italy and on other images found on internet. The model was then trained on a Tesla T4 GPU by Google Colaboratory. The trained model has produced encouraging results that can be used for real-time bubble detection
Oceans Conference Record (IEEE) - OCEANS 2021: San Diego – Porto
OCEANS 2021: San Diego – Porto
San Diego, USA
20-23 settembre 2021
Goal 9: Industry, Innovation, and Infrastructure
Bhattarai P.; Krupinski S.; Unnithan V.; Maurelli F.; Secciani N.; Franchi M.; Zacchini L.; Ridolfi A.
File in questo prodotto:
File Dimensione Formato  
A_Deep_Learning_Approach_for_Underwater_Bubble_Detection_compressed-1.pdf

Accesso chiuso

Descrizione: Articolo principale
Tipologia: Pdf editoriale (Version of record)
Licenza: DRM non definito
Dimensione 258.2 kB
Formato Adobe PDF
258.2 kB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/2158/1280644
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? ND
social impact