Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPoints
Remote sensing object detection is a basic yet challenging task in remote sensing image understanding. In contrast to horizontal objects, remote sensing objects are commonly densely packed with arbitrary orientations and highly complex backgrounds. Existing object detection methods lack an effective...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-03-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/6/1479 |
| _version_ | 1827747842249195520 |
|---|---|
| author | Lei Gao Hui Gao Yuhan Wang Dong Liu Biffon Manyura Momanyi |
| author_facet | Lei Gao Hui Gao Yuhan Wang Dong Liu Biffon Manyura Momanyi |
| author_sort | Lei Gao |
| collection | DOAJ |
| description | Remote sensing object detection is a basic yet challenging task in remote sensing image understanding. In contrast to horizontal objects, remote sensing objects are commonly densely packed with arbitrary orientations and highly complex backgrounds. Existing object detection methods lack an effective mechanism to exploit these characteristics and distinguish various targets. Unlike mainstream approaches ignoring spatial interaction among targets, this paper proposes a shape-adaptive repulsion constraint on point representation to capture geometric information of densely distributed remote sensing objects with arbitrary orientations. Specifically, (1) we first introduce a shape-adaptive center-ness quality assessment strategy to penalize the bounding boxes having a large margin shift from the center point. Then, (2) we design a novel oriented repulsion regression loss to distinguish densely packed targets: closer to the target and farther from surrounding objects. Experimental results on four challenging datasets, including DOTA, HRSC2016, UCAS-AOD, and WHU-RSONE-OBB, demonstrate the effectiveness of our proposed approach. |
| first_indexed | 2024-03-11T05:57:57Z |
| format | Article |
| id | doaj.art-188d22e78cdb43458b1e76c9ac898d48 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-11T05:57:57Z |
| publishDate | 2023-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-188d22e78cdb43458b1e76c9ac898d482023-11-17T13:37:38ZengMDPI AGRemote Sensing2072-42922023-03-01156147910.3390/rs15061479Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPointsLei Gao0Hui Gao1Yuhan Wang2Dong Liu3Biffon Manyura Momanyi4School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611700, ChinaSchool of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611700, ChinaSchool of Information and Software Engineering, University of Electronic Science and Technology of China, Chengdu 611700, ChinaSichuan Huakun Zhenyu Intelligent Technology Co., Ltd., Chengdu 610095, ChinaSchool of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611700, ChinaRemote sensing object detection is a basic yet challenging task in remote sensing image understanding. In contrast to horizontal objects, remote sensing objects are commonly densely packed with arbitrary orientations and highly complex backgrounds. Existing object detection methods lack an effective mechanism to exploit these characteristics and distinguish various targets. Unlike mainstream approaches ignoring spatial interaction among targets, this paper proposes a shape-adaptive repulsion constraint on point representation to capture geometric information of densely distributed remote sensing objects with arbitrary orientations. Specifically, (1) we first introduce a shape-adaptive center-ness quality assessment strategy to penalize the bounding boxes having a large margin shift from the center point. Then, (2) we design a novel oriented repulsion regression loss to distinguish densely packed targets: closer to the target and farther from surrounding objects. Experimental results on four challenging datasets, including DOTA, HRSC2016, UCAS-AOD, and WHU-RSONE-OBB, demonstrate the effectiveness of our proposed approach.https://www.mdpi.com/2072-4292/15/6/1479remote sensing object detectionpoint representationsample quality assessmentaerial target recognitioncenter-ness quality |
| spellingShingle | Lei Gao Hui Gao Yuhan Wang Dong Liu Biffon Manyura Momanyi Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPoints Remote Sensing remote sensing object detection point representation sample quality assessment aerial target recognition center-ness quality |
| title | Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPoints |
| title_full | Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPoints |
| title_fullStr | Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPoints |
| title_full_unstemmed | Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPoints |
| title_short | Center-Ness and Repulsion: Constraints to Improve Remote Sensing Object Detection via RepPoints |
| title_sort | center ness and repulsion constraints to improve remote sensing object detection via reppoints |
| topic | remote sensing object detection point representation sample quality assessment aerial target recognition center-ness quality |
| url | https://www.mdpi.com/2072-4292/15/6/1479 |
| work_keys_str_mv | AT leigao centernessandrepulsionconstraintstoimproveremotesensingobjectdetectionviareppoints AT huigao centernessandrepulsionconstraintstoimproveremotesensingobjectdetectionviareppoints AT yuhanwang centernessandrepulsionconstraintstoimproveremotesensingobjectdetectionviareppoints AT dongliu centernessandrepulsionconstraintstoimproveremotesensingobjectdetectionviareppoints AT biffonmanyuramomanyi centernessandrepulsionconstraintstoimproveremotesensingobjectdetectionviareppoints |