An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum Suppression
Benefiting from deep learning, synthetic aperture radar (SAR) ship detection based on convolutional neural network (CNN) has developed rapidly and corresponding performance is getting better. Nevertheless, most of the existing methods still cannot achieve a good balance between <italic>precisi...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2022-01-01
|
Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9889197/ |
_version_ | 1811204153452527616 |
---|---|
author | Chunnan Li Haitao Lang |
author_facet | Chunnan Li Haitao Lang |
author_sort | Chunnan Li |
collection | DOAJ |
description | Benefiting from deep learning, synthetic aperture radar (SAR) ship detection based on convolutional neural network (CNN) has developed rapidly and corresponding performance is getting better. Nevertheless, most of the existing methods still cannot achieve a good balance between <italic>precision</italic> and <italic>recall</italic> in scenes with complex background interferences, or in a scene where two or more ships dock side by side. To address these problems, this article proposes a novel oriented SAR ship detector, which uses oriented bounding boxes (OBBs) to describe ships. For the purpose of reducing missed ships (aiming to improve <italic>recall</italic>) while suppressing false alarms (aiming to maintain <italic>precision</italic>), the proposed detector embeds a mixed convolution channel attention (MCCA) module into the backbone network, which highlights the important feature channels to enhance ship representation features by reweighting all channels of the feature map. In addition, we consider the geometric position relationship of neighbor ships and propose geometric nonmaximum suppression (G-NMS) to remove the redundant ship candidates or possible false alarms. Extensive experiments conducted on the SSDD and HRSID<inline-formula><tex-math notation="LaTeX">$_{s}$</tex-math></inline-formula> datasets demonstrate the effectiveness of MCCA and G-NMS, the proposed detector also achieves better performance compared to state-of-the-art OBB-based detectors. |
first_indexed | 2024-04-12T03:06:38Z |
format | Article |
id | doaj.art-c1990d839a204f57b9aba7915c98ac6a |
institution | Directory Open Access Journal |
issn | 2151-1535 |
language | English |
last_indexed | 2024-04-12T03:06:38Z |
publishDate | 2022-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
spelling | doaj.art-c1990d839a204f57b9aba7915c98ac6a2022-12-22T03:50:29ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352022-01-01158074808410.1109/JSTARS.2022.32062479889197An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum SuppressionChunnan Li0https://orcid.org/0000-0002-7488-7965Haitao Lang1https://orcid.org/0000-0002-4859-1570College of Mathematics and Physics, Beijing University of Chemical Technology, Beijing, ChinaCollege of Mathematics and Physics, Beijing University of Chemical Technology, Beijing, ChinaBenefiting from deep learning, synthetic aperture radar (SAR) ship detection based on convolutional neural network (CNN) has developed rapidly and corresponding performance is getting better. Nevertheless, most of the existing methods still cannot achieve a good balance between <italic>precision</italic> and <italic>recall</italic> in scenes with complex background interferences, or in a scene where two or more ships dock side by side. To address these problems, this article proposes a novel oriented SAR ship detector, which uses oriented bounding boxes (OBBs) to describe ships. For the purpose of reducing missed ships (aiming to improve <italic>recall</italic>) while suppressing false alarms (aiming to maintain <italic>precision</italic>), the proposed detector embeds a mixed convolution channel attention (MCCA) module into the backbone network, which highlights the important feature channels to enhance ship representation features by reweighting all channels of the feature map. In addition, we consider the geometric position relationship of neighbor ships and propose geometric nonmaximum suppression (G-NMS) to remove the redundant ship candidates or possible false alarms. Extensive experiments conducted on the SSDD and HRSID<inline-formula><tex-math notation="LaTeX">$_{s}$</tex-math></inline-formula> datasets demonstrate the effectiveness of MCCA and G-NMS, the proposed detector also achieves better performance compared to state-of-the-art OBB-based detectors.https://ieeexplore.ieee.org/document/9889197/Geometric nonmaximum suppression (G-NMS)mixed convolution channel attention (MCCA)ship detectionsynthetic aperture radar (SAR) |
spellingShingle | Chunnan Li Haitao Lang An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum Suppression IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Geometric nonmaximum suppression (G-NMS) mixed convolution channel attention (MCCA) ship detection synthetic aperture radar (SAR) |
title | An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum Suppression |
title_full | An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum Suppression |
title_fullStr | An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum Suppression |
title_full_unstemmed | An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum Suppression |
title_short | An Oriented SAR Ship Detector With Mixed Convolution Channel Attention Module and Geometric Nonmaximum Suppression |
title_sort | oriented sar ship detector with mixed convolution channel attention module and geometric nonmaximum suppression |
topic | Geometric nonmaximum suppression (G-NMS) mixed convolution channel attention (MCCA) ship detection synthetic aperture radar (SAR) |
url | https://ieeexplore.ieee.org/document/9889197/ |
work_keys_str_mv | AT chunnanli anorientedsarshipdetectorwithmixedconvolutionchannelattentionmoduleandgeometricnonmaximumsuppression AT haitaolang anorientedsarshipdetectorwithmixedconvolutionchannelattentionmoduleandgeometricnonmaximumsuppression AT chunnanli orientedsarshipdetectorwithmixedconvolutionchannelattentionmoduleandgeometricnonmaximumsuppression AT haitaolang orientedsarshipdetectorwithmixedconvolutionchannelattentionmoduleandgeometricnonmaximumsuppression |