OSAD: Open-Set Aircraft Detection in SAR Images
Current mainstream synthetic aperture radar (SAR) image object detection methods still lack robustness when dealing with unknown objects in open environments. Open-set detection aims to enable detectors trained on a closed set to detect all known objects and identify unknown objects in open-set envi...
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/10815610/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832592915554107392 |
|---|---|
| author | Xiayang Xiao Zhuoxuan Li Xiaolin Mi Dandan Gu Haipeng Wang |
| author_facet | Xiayang Xiao Zhuoxuan Li Xiaolin Mi Dandan Gu Haipeng Wang |
| author_sort | Xiayang Xiao |
| collection | DOAJ |
| description | Current mainstream synthetic aperture radar (SAR) image object detection methods still lack robustness when dealing with unknown objects in open environments. Open-set detection aims to enable detectors trained on a closed set to detect all known objects and identify unknown objects in open-set environments. The key challenges are how to improve the generalization to potential unknown objects and reduce the empirical classification risk of known categories under strong supervision. To address these challenges, a novel open-set aircraft detector for SAR images is proposed, named open-set aircraft detection, which is equipped with three dedicated components: global context modeling (GCM), location quality-driven pseudolabeling generation (LPG), and prototype contrastive learning (PCL). GCM effectively enhances the network's representation of objects by attention maps that are formed by capturing long sequential positional relationships. LPG leverages clues about object positions and shapes to optimize localization quality, avoiding overfitting to known category information and enhancing generalization to potential unknown objects. PCL employs prototype-based contrastive encoding loss to promote instance-level intraclass compactness and interclass variance, aiming to minimize the overlap between known and unknown distributions and reduce the empirical classification risk of known categories. Extensive experiments have demonstrated that the proposed method can effectively detect unknown objects and exhibit competitive performance without compromising closed-set performance. The highest absolute gain that ranges from 0% to 18.36% can be achieved on the average precision of unknown objects. |
| format | Article |
| id | doaj-art-351a70bdf817428983b948128f19c520 |
| institution | Kabale University |
| 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-351a70bdf817428983b948128f19c5202025-01-21T00:00:30ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01183071308610.1109/JSTARS.2024.352224710815610OSAD: Open-Set Aircraft Detection in SAR ImagesXiayang Xiao0https://orcid.org/0000-0002-2797-7124Zhuoxuan Li1Xiaolin Mi2https://orcid.org/0009-0000-5212-9293Dandan Gu3Haipeng Wang4https://orcid.org/0000-0003-1912-7143China Mobile Internet Company, Ltd., Guangzhou, ChinaKey Laboratory for Information Science of Electromagnetic Waves (Ministry of Education), School of Information Science and Technology, Fudan University, Shanghai, ChinaKey Laboratory for Information Science of Electromagnetic Waves (Ministry of Education), School of Information Science and Technology, Fudan University, Shanghai, ChinaKey Laboratory for Information Science of Electromagnetic Waves (Ministry of Education), School of Information Science and Technology, Fudan University, Shanghai, ChinaKey Laboratory for Information Science of Electromagnetic Waves (Ministry of Education), School of Information Science and Technology, Fudan University, Shanghai, ChinaCurrent mainstream synthetic aperture radar (SAR) image object detection methods still lack robustness when dealing with unknown objects in open environments. Open-set detection aims to enable detectors trained on a closed set to detect all known objects and identify unknown objects in open-set environments. The key challenges are how to improve the generalization to potential unknown objects and reduce the empirical classification risk of known categories under strong supervision. To address these challenges, a novel open-set aircraft detector for SAR images is proposed, named open-set aircraft detection, which is equipped with three dedicated components: global context modeling (GCM), location quality-driven pseudolabeling generation (LPG), and prototype contrastive learning (PCL). GCM effectively enhances the network's representation of objects by attention maps that are formed by capturing long sequential positional relationships. LPG leverages clues about object positions and shapes to optimize localization quality, avoiding overfitting to known category information and enhancing generalization to potential unknown objects. PCL employs prototype-based contrastive encoding loss to promote instance-level intraclass compactness and interclass variance, aiming to minimize the overlap between known and unknown distributions and reduce the empirical classification risk of known categories. Extensive experiments have demonstrated that the proposed method can effectively detect unknown objects and exhibit competitive performance without compromising closed-set performance. The highest absolute gain that ranges from 0% to 18.36% can be achieved on the average precision of unknown objects.https://ieeexplore.ieee.org/document/10815610/Context modelingconvolutional neural network (CNN)open-set detectionprototype learningsynthetic aperture radar (SAR) |
| spellingShingle | Xiayang Xiao Zhuoxuan Li Xiaolin Mi Dandan Gu Haipeng Wang OSAD: Open-Set Aircraft Detection in SAR Images IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Context modeling convolutional neural network (CNN) open-set detection prototype learning synthetic aperture radar (SAR) |
| title | OSAD: Open-Set Aircraft Detection in SAR Images |
| title_full | OSAD: Open-Set Aircraft Detection in SAR Images |
| title_fullStr | OSAD: Open-Set Aircraft Detection in SAR Images |
| title_full_unstemmed | OSAD: Open-Set Aircraft Detection in SAR Images |
| title_short | OSAD: Open-Set Aircraft Detection in SAR Images |
| title_sort | osad open set aircraft detection in sar images |
| topic | Context modeling convolutional neural network (CNN) open-set detection prototype learning synthetic aperture radar (SAR) |
| url | https://ieeexplore.ieee.org/document/10815610/ |
| work_keys_str_mv | AT xiayangxiao osadopensetaircraftdetectioninsarimages AT zhuoxuanli osadopensetaircraftdetectioninsarimages AT xiaolinmi osadopensetaircraftdetectioninsarimages AT dandangu osadopensetaircraftdetectioninsarimages AT haipengwang osadopensetaircraftdetectioninsarimages |