FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR image
Heterogeneous Synthetic Aperture Radar (SAR) image object detection task with inconsistent joint probability distributions is occurring more and more frequently in practical applications. In which the small sample of data scarcity is becoming an urgent problem for researchers. Therefore, this paper...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-06-01
|
| Series: | Science of Remote Sensing |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2666017225000082 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849689199647653888 |
|---|---|
| author | Siyuan Zhao Yong Kang Hang Yuan Guan Wang Hui Wang Shichao Xiong Ying Luo |
| author_facet | Siyuan Zhao Yong Kang Hang Yuan Guan Wang Hui Wang Shichao Xiong Ying Luo |
| author_sort | Siyuan Zhao |
| collection | DOAJ |
| description | Heterogeneous Synthetic Aperture Radar (SAR) image object detection task with inconsistent joint probability distributions is occurring more and more frequently in practical applications. In which the small sample of data scarcity is becoming an urgent problem for researchers. Therefore, this paper proposes a novel few-shot domain adaptation object detection (FsDAOD) method based on Faster Region Convolutional Neural Network baseline to cope with the above problem. Firstly, employing the foundational structure of the existing baseline method, a novel mutual information loss function is introduced that prompts the neural network to extract domain-specific knowledge. This strategic approach encourages distinctive levels of confidence in individual predictions while fostering overall diversity. Given that performance can be easily over-fitted with a restricted number of observed objects if feature alignment strictly adheres to conventional methods, the set of source instances are initially categorized into two groups: target domain-easy set and target domain-hard set. Subsequently, asynchronous alignment is performed between the target-hard domain set of the source instances and the extended dataset of the target instances to achieve effective supervised learning. It is then asserted that confidence-based sample separation methods can improve detection efficiency by adjusting the model to prioritize the identification of more easily detected objects, but this may lead to incorrect decisions for more challenging instances. Extensive experiments on FsDAOD on heterogeneous satellite-borne SAR image datasets have been conducted, and the experimental results have demonstrated that the detection rate of the proposed method exceeds the existing state-of-the-art methods by 5%. |
| format | Article |
| id | doaj-art-ae3da9158e1a4beeae6309f9c7d96928 |
| institution | DOAJ |
| issn | 2666-0172 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Science of Remote Sensing |
| spelling | doaj-art-ae3da9158e1a4beeae6309f9c7d969282025-08-20T03:21:43ZengElsevierScience of Remote Sensing2666-01722025-06-011110020210.1016/j.srs.2025.100202FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR imageSiyuan Zhao0Yong Kang1Hang Yuan2Guan Wang3Hui Wang4Shichao Xiong5Ying Luo6Comprehensive Training Department, Air Force Communication NCO Academy, Dalian 116000, China; Institute of Information and Navigation, Air Force Engineering University, Xi’an 710077, China; Corresponding author at: Comprehensive Training Department, Air Force Communication NCO Academy, Dalian 116000, China.Comprehensive Training Department, Air Force Communication NCO Academy, Dalian 116000, ChinaInstitute of Information and Navigation, Air Force Engineering University, Xi’an 710077, ChinaComprehensive Training Department, Air Force Communication NCO Academy, Dalian 116000, ChinaComprehensive Training Department, Air Force Communication NCO Academy, Dalian 116000, ChinaInstitute of Information and Navigation, Air Force Engineering University, Xi’an 710077, ChinaInstitute of Information and Navigation, Air Force Engineering University, Xi’an 710077, China; Collaborative Innovation Center of Information Sensing and Understanding, Xi’an 710077, ChinaHeterogeneous Synthetic Aperture Radar (SAR) image object detection task with inconsistent joint probability distributions is occurring more and more frequently in practical applications. In which the small sample of data scarcity is becoming an urgent problem for researchers. Therefore, this paper proposes a novel few-shot domain adaptation object detection (FsDAOD) method based on Faster Region Convolutional Neural Network baseline to cope with the above problem. Firstly, employing the foundational structure of the existing baseline method, a novel mutual information loss function is introduced that prompts the neural network to extract domain-specific knowledge. This strategic approach encourages distinctive levels of confidence in individual predictions while fostering overall diversity. Given that performance can be easily over-fitted with a restricted number of observed objects if feature alignment strictly adheres to conventional methods, the set of source instances are initially categorized into two groups: target domain-easy set and target domain-hard set. Subsequently, asynchronous alignment is performed between the target-hard domain set of the source instances and the extended dataset of the target instances to achieve effective supervised learning. It is then asserted that confidence-based sample separation methods can improve detection efficiency by adjusting the model to prioritize the identification of more easily detected objects, but this may lead to incorrect decisions for more challenging instances. Extensive experiments on FsDAOD on heterogeneous satellite-borne SAR image datasets have been conducted, and the experimental results have demonstrated that the detection rate of the proposed method exceeds the existing state-of-the-art methods by 5%.http://www.sciencedirect.com/science/article/pii/S2666017225000082Heterogeneous Synthetic Aperture Radar (SAR) imagesFew-shot domain adaptationMutual informationSample separation |
| spellingShingle | Siyuan Zhao Yong Kang Hang Yuan Guan Wang Hui Wang Shichao Xiong Ying Luo FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR image Science of Remote Sensing Heterogeneous Synthetic Aperture Radar (SAR) images Few-shot domain adaptation Mutual information Sample separation |
| title | FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR image |
| title_full | FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR image |
| title_fullStr | FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR image |
| title_full_unstemmed | FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR image |
| title_short | FsDAOD: Few-shot domain adaptation object detection for heterogeneous SAR image |
| title_sort | fsdaod few shot domain adaptation object detection for heterogeneous sar image |
| topic | Heterogeneous Synthetic Aperture Radar (SAR) images Few-shot domain adaptation Mutual information Sample separation |
| url | http://www.sciencedirect.com/science/article/pii/S2666017225000082 |
| work_keys_str_mv | AT siyuanzhao fsdaodfewshotdomainadaptationobjectdetectionforheterogeneoussarimage AT yongkang fsdaodfewshotdomainadaptationobjectdetectionforheterogeneoussarimage AT hangyuan fsdaodfewshotdomainadaptationobjectdetectionforheterogeneoussarimage AT guanwang fsdaodfewshotdomainadaptationobjectdetectionforheterogeneoussarimage AT huiwang fsdaodfewshotdomainadaptationobjectdetectionforheterogeneoussarimage AT shichaoxiong fsdaodfewshotdomainadaptationobjectdetectionforheterogeneoussarimage AT yingluo fsdaodfewshotdomainadaptationobjectdetectionforheterogeneoussarimage |