D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime

Yoann Arhant; Olga Lopera Tellez; Xavier Neyt; Aleksandra Pizurica

doi:10.1109/IGARSS52108.2023.10282903

D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime

Yoann Arhant, Olga Lopera Tellez, Xavier Neyt, Aleksandra Pizurica

University of Ghent

Publikation: Beitrag in Buch/Bericht/Konferenzband › Konferenzbeitrag › Begutachtung

3 Downloads (Pure)

Abstract

Seabed characterisation consists in the study of the physical and biological properties of the bottom of the oceans. It is effectively achieved with sonar, a remote sensing method that captures acoustic backscatter of the seabed. Classical Machine Learning (ML) and Deep Learning (DL) research have failed to successfully address the automatic mapping of the seabed from noisy sonar data. This work introduces the Deep Supervised Semantic Segmentation model for Seabed Characterisation (D4SC), a novel U-Net-like model tailored to such data and low-label regime, and proposes a new end-to-end processing pipeline for seabed semantic segmentation. That dual contribution achieves state-of-the-art results on a high resolution Synthetic Aperture Sonar (SAS) survey dataset.

Originalsprache	Englisch
Titel	IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings
Herausgeber (Verlag)	Institute of Electrical and Electronics Engineers Inc.
Seiten	6932-6935
Seitenumfang	4
ISBN (elektronisch)	9798350320107
DOIs	https://doi.org/10.1109/IGARSS52108.2023.10282903
Publikationsstatus	Veröffentlicht - 2023
Veranstaltung	2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023 - Pasadena, USA/Vereinigte Staaten Dauer: 16 Juli 2023 → 21 Juli 2023 https://2023.ieeeigarss.org/

Publikationsreihe

Name	International Geoscience and Remote Sensing Symposium (IGARSS)
Band	2023-July

Konferenz

Konferenz	2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023
Kurztitel	IGARSS 2023
Land/Gebiet	USA/Vereinigte Staaten
Ort	Pasadena
Zeitraum	16/07/23 → 21/07/23
Internetadresse	https://2023.ieeeigarss.org/

Zugriff auf Dokument

10.1109/IGARSS52108.2023.10282903

preprintD4SCAkzeptiertes Autorenmanuskript, 4,67 MBLizenz: CC BY-NC-ND

Andere Dateien und Links

Verknüpfung zur Publikation in Scopus

DAP/21-11: Automatic, small-scale sea-floor characterization from high-resolution sonar data
Lopera Tellez, O. (Leitende(r) Forscher/-in), Neyt, X. (Leitende(r) Forscher/-in) & Arhant, Y. (Forschende)
1/09/21 → 31/08/25
Projekt: Forschung › Research

Dieses zitieren

Arhant, Y., Tellez, O. L., Neyt, X., & Pizurica, A. (2023). D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime. in IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings (S. 6932-6935). (International Geoscience and Remote Sensing Symposium (IGARSS); Band 2023-July). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IGARSS52108.2023.10282903

Arhant, Yoann ; Tellez, Olga Lopera ; Neyt, Xavier et al. / D4SC : Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime. IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2023. S. 6932-6935 (International Geoscience and Remote Sensing Symposium (IGARSS)).

@inproceedings{fa9acf268fab4efca1416a1524d3e369,

title = "D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime",

abstract = "Seabed characterisation consists in the study of the physical and biological properties of the bottom of the oceans. It is effectively achieved with sonar, a remote sensing method that captures acoustic backscatter of the seabed. Classical Machine Learning (ML) and Deep Learning (DL) research have failed to successfully address the automatic mapping of the seabed from noisy sonar data. This work introduces the Deep Supervised Semantic Segmentation model for Seabed Characterisation (D4SC), a novel U-Net-like model tailored to such data and low-label regime, and proposes a new end-to-end processing pipeline for seabed semantic segmentation. That dual contribution achieves state-of-the-art results on a high resolution Synthetic Aperture Sonar (SAS) survey dataset.",

keywords = "Deep Learning, Seabed, Semantic Segmentation, Synthetic Aperture Sonar",

author = "Yoann Arhant and Tellez, {Olga Lopera} and Xavier Neyt and Aleksandra Pizurica",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023, IGARSS 2023 ; Conference date: 16-07-2023 Through 21-07-2023",

year = "2023",

doi = "10.1109/IGARSS52108.2023.10282903",

language = "English",

series = "International Geoscience and Remote Sensing Symposium (IGARSS)",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "6932--6935",

booktitle = "IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings",

url = "https://2023.ieeeigarss.org/",

}

Arhant, Y , Tellez, OL , Neyt, X & Pizurica, A 2023, D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime. in IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. International Geoscience and Remote Sensing Symposium (IGARSS), Bd. 2023-July, Institute of Electrical and Electronics Engineers Inc., S. 6932-6935, 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023, Pasadena, USA/Vereinigte Staaten, 16/07/23. https://doi.org/10.1109/IGARSS52108.2023.10282903

D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime. / Arhant, Yoann ; Tellez, Olga Lopera ; Neyt, Xavier et al.
IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2023. S. 6932-6935 (International Geoscience and Remote Sensing Symposium (IGARSS); Band 2023-July).

Publikation: Beitrag in Buch/Bericht/Konferenzband › Konferenzbeitrag › Begutachtung

TY - GEN

T1 - D4SC

T2 - 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023

AU - Arhant, Yoann

AU - Tellez, Olga Lopera

AU - Neyt, Xavier

AU - Pizurica, Aleksandra

PY - 2023

Y1 - 2023

N2 - Seabed characterisation consists in the study of the physical and biological properties of the bottom of the oceans. It is effectively achieved with sonar, a remote sensing method that captures acoustic backscatter of the seabed. Classical Machine Learning (ML) and Deep Learning (DL) research have failed to successfully address the automatic mapping of the seabed from noisy sonar data. This work introduces the Deep Supervised Semantic Segmentation model for Seabed Characterisation (D4SC), a novel U-Net-like model tailored to such data and low-label regime, and proposes a new end-to-end processing pipeline for seabed semantic segmentation. That dual contribution achieves state-of-the-art results on a high resolution Synthetic Aperture Sonar (SAS) survey dataset.

AB - Seabed characterisation consists in the study of the physical and biological properties of the bottom of the oceans. It is effectively achieved with sonar, a remote sensing method that captures acoustic backscatter of the seabed. Classical Machine Learning (ML) and Deep Learning (DL) research have failed to successfully address the automatic mapping of the seabed from noisy sonar data. This work introduces the Deep Supervised Semantic Segmentation model for Seabed Characterisation (D4SC), a novel U-Net-like model tailored to such data and low-label regime, and proposes a new end-to-end processing pipeline for seabed semantic segmentation. That dual contribution achieves state-of-the-art results on a high resolution Synthetic Aperture Sonar (SAS) survey dataset.

KW - Deep Learning

KW - Seabed

KW - Semantic Segmentation

KW - Synthetic Aperture Sonar

UR - http://www.scopus.com/inward/record.url?scp=85178379391&partnerID=8YFLogxK

U2 - 10.1109/IGARSS52108.2023.10282903

DO - 10.1109/IGARSS52108.2023.10282903

M3 - Conference contribution

AN - SCOPUS:85178379391

T3 - International Geoscience and Remote Sensing Symposium (IGARSS)

SP - 6932

EP - 6935

BT - IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 16 July 2023 through 21 July 2023

ER -

Arhant Y , Tellez OL , Neyt X, Pizurica A. D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime. in IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. Institute of Electrical and Electronics Engineers Inc. 2023. S. 6932-6935. (International Geoscience and Remote Sensing Symposium (IGARSS)). doi: 10.1109/IGARSS52108.2023.10282903

D4SC: Deep Supervised Semantic Segmentation for Seabed Characterisation in Low-Label Regime

Abstract

Publikationsreihe

Konferenz

Zugriff auf Dokument

Andere Dateien und Links

Fingerprint

Projekte

DAP/21-11: Automatic, small-scale sea-floor characterization from high-resolution sonar data

Dieses zitieren