R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable Proposals
Weintroduce R-Sparse R-CNN, a novel pipeline for oriented ship detection in Synthetic Aperture Radar (SAR) images that leverages sparse learnable proposals enriched with background contextual information, termed background-aware proposals (BAPs). The adoption of sparse proposals streamlines the pipe...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11027781/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1839648898763718656 |
|---|---|
| author | Kamirul Kamirul Odysseas A. Pappas Alin M. Achim |
| author_facet | Kamirul Kamirul Odysseas A. Pappas Alin M. Achim |
| author_sort | Kamirul Kamirul |
| collection | DOAJ |
| description | Weintroduce R-Sparse R-CNN, a novel pipeline for oriented ship detection in Synthetic Aperture Radar (SAR) images that leverages sparse learnable proposals enriched with background contextual information, termed background-aware proposals (BAPs). The adoption of sparse proposals streamlines the pipeline by eliminating the need for proposal generators and post-processing for overlapping predictions. The proposed BAPs enrich object representation by integrating ship and background features, allowing the model to learn their contextual relationships for more accurate distinction of ships in complex environments. To complement BAPs, we propose dual-context pooling (DCP), a novel strategy that jointly extracts ship and background features in a single unified operation. This unified design improves efficiency by eliminating redundant computation inherent in separate pooling. Moreover, by ensuring that ship and background features are pooled from the same feature map level, DCP provides aligned features that improve contextual relationship learning. Finally, as a core component of contextual relationship learning in R-Sparse R-CNN, we design a dedicated transformer-based interaction module. This module interacts pooled ship and background features with corresponding proposal features and models their relationships. Experimental results show that R-Sparse R-CNN delivers outstanding accuracy, surpassing state-of-the-art models by margins of up to 12.8% and 11.9% on SSDD and RSDD-SAR inshore datasets, respectively. These results demonstrate the effectiveness and competitiveness of R-Sparse R-CNN as a robust framework for oriented ship detection in SAR imagery. |
| format | Article |
| id | doaj-art-941569e0158748689e074f5f5ea68c38 |
| institution | Matheson Library |
| issn | 1939-1404 2151-1535 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj-art-941569e0158748689e074f5f5ea68c382025-06-27T23:00:23ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118149551497310.1109/JSTARS.2025.357776611027781R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable ProposalsKamirul Kamirul0https://orcid.org/0000-0002-1474-1139Odysseas A. Pappas1https://orcid.org/0000-0003-3037-2828Alin M. Achim2https://orcid.org/0000-0002-0982-7798Visual Information Laboratory, University of Bristol, Bristol, U.K.Visual Information Laboratory, University of Bristol, Bristol, U.K.Visual Information Laboratory, University of Bristol, Bristol, U.K.Weintroduce R-Sparse R-CNN, a novel pipeline for oriented ship detection in Synthetic Aperture Radar (SAR) images that leverages sparse learnable proposals enriched with background contextual information, termed background-aware proposals (BAPs). The adoption of sparse proposals streamlines the pipeline by eliminating the need for proposal generators and post-processing for overlapping predictions. The proposed BAPs enrich object representation by integrating ship and background features, allowing the model to learn their contextual relationships for more accurate distinction of ships in complex environments. To complement BAPs, we propose dual-context pooling (DCP), a novel strategy that jointly extracts ship and background features in a single unified operation. This unified design improves efficiency by eliminating redundant computation inherent in separate pooling. Moreover, by ensuring that ship and background features are pooled from the same feature map level, DCP provides aligned features that improve contextual relationship learning. Finally, as a core component of contextual relationship learning in R-Sparse R-CNN, we design a dedicated transformer-based interaction module. This module interacts pooled ship and background features with corresponding proposal features and models their relationships. Experimental results show that R-Sparse R-CNN delivers outstanding accuracy, surpassing state-of-the-art models by margins of up to 12.8% and 11.9% on SSDD and RSDD-SAR inshore datasets, respectively. These results demonstrate the effectiveness and competitiveness of R-Sparse R-CNN as a robust framework for oriented ship detection in SAR imagery.https://ieeexplore.ieee.org/document/11027781/Background-aware proposalsconvolutional neural networksdeep learningsparse learnable proposalssynthetic aperture radaroriented ship detection |
| spellingShingle | Kamirul Kamirul Odysseas A. Pappas Alin M. Achim R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable Proposals IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Background-aware proposals convolutional neural networks deep learning sparse learnable proposals synthetic aperture radar oriented ship detection |
| title | R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable Proposals |
| title_full | R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable Proposals |
| title_fullStr | R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable Proposals |
| title_full_unstemmed | R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable Proposals |
| title_short | R-Sparse R-CNN: SAR Ship Detection Based on Background-Aware Sparse Learnable Proposals |
| title_sort | r sparse r cnn sar ship detection based on background aware sparse learnable proposals |
| topic | Background-aware proposals convolutional neural networks deep learning sparse learnable proposals synthetic aperture radar oriented ship detection |
| url | https://ieeexplore.ieee.org/document/11027781/ |
| work_keys_str_mv | AT kamirulkamirul rsparsercnnsarshipdetectionbasedonbackgroundawaresparselearnableproposals AT odysseasapappas rsparsercnnsarshipdetectionbasedonbackgroundawaresparselearnableproposals AT alinmachim rsparsercnnsarshipdetectionbasedonbackgroundawaresparselearnableproposals |