An Improved Lightweight RetinaNet for Ship Detection in SAR Images
The rapid development of remote sensing technology has led to a sharp increase in the amount of synthetic aperture radar (SAR) measurements, which put forward higher requirements for remote sensing image processing. As an important application of SAR, fast and accurate ship detection has always been...
| 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/9788025/ |
| _version_ | 1818550785886126080 |
|---|---|
| author | Tian Miao HongCheng Zeng Wei Yang Boce Chu Fei Zou Weijia Ren Jie Chen |
| author_facet | Tian Miao HongCheng Zeng Wei Yang Boce Chu Fei Zou Weijia Ren Jie Chen |
| author_sort | Tian Miao |
| collection | DOAJ |
| description | The rapid development of remote sensing technology has led to a sharp increase in the amount of synthetic aperture radar (SAR) measurements, which put forward higher requirements for remote sensing image processing. As an important application of SAR, fast and accurate ship detection has always been a research hotspot. In this article, an improved lightweight RetinaNet for ship detection in SAR images is proposed. Compared with the standard RetinaNet, the shallow convolutional layers of the backbone are replaced by ghost modules and the number of the deep convolutional layers is reduced. The spatial and channel attention modules are embedded into the model to enhance detectability. <italic>K</italic>-means clustering algorithm is applied to adjust the initial aspect ratios of the model. The effectiveness and robustness of the proposed method is demonstrated by numerical experiments with SSDD dataset, Gaofen-3 mini dataset, and a large-scale SAR image of Hisea-1 satellite, it is shown that the proposed method can significantly reduce the floating-point operations and the number of parameters without decreasing the detection accuracy and recall ratio. Moreover, the experimental results also show the proposed model's robustness and the ability to detect ship targets in small datasets. |
| first_indexed | 2024-12-12T08:51:08Z |
| format | Article |
| id | doaj.art-551968674c1940e6a7fecd923c20b59f |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-12-12T08:51:08Z |
| 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-551968674c1940e6a7fecd923c20b59f2022-12-22T00:30:12ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352022-01-01154667467910.1109/JSTARS.2022.31801599788025An Improved Lightweight RetinaNet for Ship Detection in SAR ImagesTian Miao0https://orcid.org/0000-0002-5888-2064HongCheng Zeng1https://orcid.org/0000-0002-4727-9980Wei Yang2https://orcid.org/0000-0001-8935-294XBoce Chu3https://orcid.org/0000-0002-5231-0620Fei Zou4Weijia Ren5Jie Chen6https://orcid.org/0000-0002-9370-3965School of Electronic and Information Engineering, Beihang University, Beijing, ChinaSchool of Electronic and Information Engineering, Beihang University, Beijing, ChinaSchool of Electronic and Information Engineering, Beihang University, Beijing, ChinaCETC Key Laboratory of Aerospace Information Applications, Shijiazhuang, ChinaBeijing Institute of Remote Sensing Information, Beijing, ChinaSpacety Company, Ltd. (Changsha), Changsha, ChinaSchool of Electronic and Information Engineering, Beihang University, Beijing, ChinaThe rapid development of remote sensing technology has led to a sharp increase in the amount of synthetic aperture radar (SAR) measurements, which put forward higher requirements for remote sensing image processing. As an important application of SAR, fast and accurate ship detection has always been a research hotspot. In this article, an improved lightweight RetinaNet for ship detection in SAR images is proposed. Compared with the standard RetinaNet, the shallow convolutional layers of the backbone are replaced by ghost modules and the number of the deep convolutional layers is reduced. The spatial and channel attention modules are embedded into the model to enhance detectability. <italic>K</italic>-means clustering algorithm is applied to adjust the initial aspect ratios of the model. The effectiveness and robustness of the proposed method is demonstrated by numerical experiments with SSDD dataset, Gaofen-3 mini dataset, and a large-scale SAR image of Hisea-1 satellite, it is shown that the proposed method can significantly reduce the floating-point operations and the number of parameters without decreasing the detection accuracy and recall ratio. Moreover, the experimental results also show the proposed model's robustness and the ability to detect ship targets in small datasets.https://ieeexplore.ieee.org/document/9788025/Lightweight designRetinanetsynthetic aperture radar (SAR)ship detection |
| spellingShingle | Tian Miao HongCheng Zeng Wei Yang Boce Chu Fei Zou Weijia Ren Jie Chen An Improved Lightweight RetinaNet for Ship Detection in SAR Images IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Lightweight design Retinanet synthetic aperture radar (SAR) ship detection |
| title | An Improved Lightweight RetinaNet for Ship Detection in SAR Images |
| title_full | An Improved Lightweight RetinaNet for Ship Detection in SAR Images |
| title_fullStr | An Improved Lightweight RetinaNet for Ship Detection in SAR Images |
| title_full_unstemmed | An Improved Lightweight RetinaNet for Ship Detection in SAR Images |
| title_short | An Improved Lightweight RetinaNet for Ship Detection in SAR Images |
| title_sort | improved lightweight retinanet for ship detection in sar images |
| topic | Lightweight design Retinanet synthetic aperture radar (SAR) ship detection |
| url | https://ieeexplore.ieee.org/document/9788025/ |
| work_keys_str_mv | AT tianmiao animprovedlightweightretinanetforshipdetectioninsarimages AT hongchengzeng animprovedlightweightretinanetforshipdetectioninsarimages AT weiyang animprovedlightweightretinanetforshipdetectioninsarimages AT bocechu animprovedlightweightretinanetforshipdetectioninsarimages AT feizou animprovedlightweightretinanetforshipdetectioninsarimages AT weijiaren animprovedlightweightretinanetforshipdetectioninsarimages AT jiechen animprovedlightweightretinanetforshipdetectioninsarimages AT tianmiao improvedlightweightretinanetforshipdetectioninsarimages AT hongchengzeng improvedlightweightretinanetforshipdetectioninsarimages AT weiyang improvedlightweightretinanetforshipdetectioninsarimages AT bocechu improvedlightweightretinanetforshipdetectioninsarimages AT feizou improvedlightweightretinanetforshipdetectioninsarimages AT weijiaren improvedlightweightretinanetforshipdetectioninsarimages AT jiechen improvedlightweightretinanetforshipdetectioninsarimages |