Deep Cascade Network for Noise-Robust SAR Ship Detection With Label Augmentation

Keunhoon Choi; Taeyong Song; Sunok Kim; Hyunsung Jang; Namkoo Ha; Kwanghoon Sohn

doi:10.1109/LGRS.2022.3205715

Deep Cascade Network for Noise-Robust SAR Ship Detection With Label Augmentation

Keunhoon Choi, Taeyong Song, Sunok Kim, Hyunsung Jang, Namkoo Ha, Kwanghoon Sohn

Department of Electrical and Electronic Engineering

Research output: Contribution to journal › Article › peer-review

Abstract

Deep learning has recently made an impressive advance in ship detection in Synthetic Aperture Radar (SAR) images. Despite this advancement, conventional deep detection networks often suffer from speckle noise that inherently occurs in SAR images. However, despeckling researches have focused only on improving the visual quality of the SAR images. Despeckling without considering subsequent task may cause loss of semantic information and result in performance degradation. In this letter, we propose a deep cascade framework for noise-robust SAR ship detection that sequentially performs despeckle and detection. We effectively train our cascade network using pseudo-SAR images with SAR-like structures and additional detection annotations. We also propose semantic conservative loss that allows these two tasks to cooperate with each other. Experimental results including comparisons to previous methods and extensive ablation studies show the effectiveness of our proposed method.

Original language	English
Article number	4514005
Journal	IEEE Geoscience and Remote Sensing Letters
Volume	19
DOIs	https://doi.org/10.1109/LGRS.2022.3205715
Publication status	Published - 2022

Bibliographical note

Funding Information:
This work was supported in part by the Institute of Information and Communications Technology Planning and Evaluation (IITP) Grant by the Korean Government (MSIT) (Artificial Intelligence Innovation Hub) under Grant 2021-0-02068 and in part by the Yonsei University Research Fund of 2022 under Grant 2022-22-0002. The work of Sunok Kim was supported by the National Research Foundation of Korea (NRF) Grant by the Korean Government (MSIP) under Grant NRF-2021R1C1C2005202.

Publisher Copyright:
© 2004-2012 IEEE.

All Science Journal Classification (ASJC) codes

Geotechnical Engineering and Engineering Geology
Electrical and Electronic Engineering

Access to Document

10.1109/LGRS.2022.3205715

Cite this

@article{a3e2de6825bc4f2b95a838d3b70cd95a,

title = "Deep Cascade Network for Noise-Robust SAR Ship Detection With Label Augmentation",

abstract = "Deep learning has recently made an impressive advance in ship detection in Synthetic Aperture Radar (SAR) images. Despite this advancement, conventional deep detection networks often suffer from speckle noise that inherently occurs in SAR images. However, despeckling researches have focused only on improving the visual quality of the SAR images. Despeckling without considering subsequent task may cause loss of semantic information and result in performance degradation. In this letter, we propose a deep cascade framework for noise-robust SAR ship detection that sequentially performs despeckle and detection. We effectively train our cascade network using pseudo-SAR images with SAR-like structures and additional detection annotations. We also propose semantic conservative loss that allows these two tasks to cooperate with each other. Experimental results including comparisons to previous methods and extensive ablation studies show the effectiveness of our proposed method.",

author = "Keunhoon Choi and Taeyong Song and Sunok Kim and Hyunsung Jang and Namkoo Ha and Kwanghoon Sohn",

note = "Funding Information: This work was supported in part by the Institute of Information and Communications Technology Planning and Evaluation (IITP) Grant by the Korean Government (MSIT) (Artificial Intelligence Innovation Hub) under Grant 2021-0-02068 and in part by the Yonsei University Research Fund of 2022 under Grant 2022-22-0002. The work of Sunok Kim was supported by the National Research Foundation of Korea (NRF) Grant by the Korean Government (MSIP) under Grant NRF-2021R1C1C2005202. Publisher Copyright: {\textcopyright} 2004-2012 IEEE.",

year = "2022",

doi = "10.1109/LGRS.2022.3205715",

language = "English",

volume = "19",

journal = "IEEE Geoscience and Remote Sensing Letters",

issn = "1545-598X",

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

}

TY - JOUR

T1 - Deep Cascade Network for Noise-Robust SAR Ship Detection With Label Augmentation

AU - Choi, Keunhoon

AU - Song, Taeyong

AU - Kim, Sunok

AU - Jang, Hyunsung

AU - Ha, Namkoo

AU - Sohn, Kwanghoon

N1 - Funding Information: This work was supported in part by the Institute of Information and Communications Technology Planning and Evaluation (IITP) Grant by the Korean Government (MSIT) (Artificial Intelligence Innovation Hub) under Grant 2021-0-02068 and in part by the Yonsei University Research Fund of 2022 under Grant 2022-22-0002. The work of Sunok Kim was supported by the National Research Foundation of Korea (NRF) Grant by the Korean Government (MSIP) under Grant NRF-2021R1C1C2005202. Publisher Copyright: © 2004-2012 IEEE.

PY - 2022

Y1 - 2022

N2 - Deep learning has recently made an impressive advance in ship detection in Synthetic Aperture Radar (SAR) images. Despite this advancement, conventional deep detection networks often suffer from speckle noise that inherently occurs in SAR images. However, despeckling researches have focused only on improving the visual quality of the SAR images. Despeckling without considering subsequent task may cause loss of semantic information and result in performance degradation. In this letter, we propose a deep cascade framework for noise-robust SAR ship detection that sequentially performs despeckle and detection. We effectively train our cascade network using pseudo-SAR images with SAR-like structures and additional detection annotations. We also propose semantic conservative loss that allows these two tasks to cooperate with each other. Experimental results including comparisons to previous methods and extensive ablation studies show the effectiveness of our proposed method.

AB - Deep learning has recently made an impressive advance in ship detection in Synthetic Aperture Radar (SAR) images. Despite this advancement, conventional deep detection networks often suffer from speckle noise that inherently occurs in SAR images. However, despeckling researches have focused only on improving the visual quality of the SAR images. Despeckling without considering subsequent task may cause loss of semantic information and result in performance degradation. In this letter, we propose a deep cascade framework for noise-robust SAR ship detection that sequentially performs despeckle and detection. We effectively train our cascade network using pseudo-SAR images with SAR-like structures and additional detection annotations. We also propose semantic conservative loss that allows these two tasks to cooperate with each other. Experimental results including comparisons to previous methods and extensive ablation studies show the effectiveness of our proposed method.

UR - http://www.scopus.com/inward/record.url?scp=85139444720&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85139444720&partnerID=8YFLogxK

U2 - 10.1109/LGRS.2022.3205715

DO - 10.1109/LGRS.2022.3205715

M3 - Article

AN - SCOPUS:85139444720

SN - 1545-598X

VL - 19

JO - IEEE Geoscience and Remote Sensing Letters

JF - IEEE Geoscience and Remote Sensing Letters

M1 - 4514005

ER -

Deep Cascade Network for Noise-Robust SAR Ship Detection With Label Augmentation

Abstract

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this