With the specific aim of improving our Maritime Domain Awareness, satellite data enable a wide range of applications, including fisheries and pollution control, anti-piracy actions, and surveillance over coastal/protected regions. Among all the available data, the ones gathered by space-borne synthetic aperture radar (SAR) are attracting large interest thanks to their coverage and all-weather and all-Time observation capabilities. Currently, Artificial Intelligence (AI) has been widely recognized as the only way to take fully advantages of increasing amount of Earth Observation (EO) data, and Deep Learning-based detectors have been successfully applied for the detection of ships from cluttered sea surface. However, nonetheless their exploitation for ship route estimation purposes, the problem of wake detection by deep learning has been barely touched. With this concern, the paper investigates one of the latest deep learning architecture for object detection, i.e. RetinaNet, as an effective means to achieve a robust wake detector.

RetinaNet: A deep learning architecture to achieve a robust wake detector in SAR images / Del Prete, R.; Graziano, M. D.; Renga, A.. - (2021), pp. 171-176. (Intervento presentato al convegno 6th International Forum on Research and Technology for Society and Industry, RTSI 2021 tenutosi a ita nel 2021) [10.1109/RTSI50628.2021.9597297].

RetinaNet: A deep learning architecture to achieve a robust wake detector in SAR images

Del Prete R.;Graziano M. D.;Renga A.
2021

Abstract

With the specific aim of improving our Maritime Domain Awareness, satellite data enable a wide range of applications, including fisheries and pollution control, anti-piracy actions, and surveillance over coastal/protected regions. Among all the available data, the ones gathered by space-borne synthetic aperture radar (SAR) are attracting large interest thanks to their coverage and all-weather and all-Time observation capabilities. Currently, Artificial Intelligence (AI) has been widely recognized as the only way to take fully advantages of increasing amount of Earth Observation (EO) data, and Deep Learning-based detectors have been successfully applied for the detection of ships from cluttered sea surface. However, nonetheless their exploitation for ship route estimation purposes, the problem of wake detection by deep learning has been barely touched. With this concern, the paper investigates one of the latest deep learning architecture for object detection, i.e. RetinaNet, as an effective means to achieve a robust wake detector.
2021
978-1-6654-4135-3
RetinaNet: A deep learning architecture to achieve a robust wake detector in SAR images / Del Prete, R.; Graziano, M. D.; Renga, A.. - (2021), pp. 171-176. (Intervento presentato al convegno 6th International Forum on Research and Technology for Society and Industry, RTSI 2021 tenutosi a ita nel 2021) [10.1109/RTSI50628.2021.9597297].
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS 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: https://hdl.handle.net/11588/892204
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? ND
social impact