Oriented object detection in aerial images with box boundary-aware vectors. Yi, J., Wu, P., Liu, B., Huang, Q., Qu, H., & Metaxas, D. Proceedings - 2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021, 2021. ![Oriented object detection in aerial images with box boundary-aware vectors [pdf]](https://bibbase.org/img/filetypes/pdf.svg "pdf")
Paper doi abstract bibtex Oriented object detection in aerial images is a challenging task as the objects in aerial images are displayed in arbitrary directions and are usually densely packed. Cur-rent oriented object detection methods mainly rely on two-stage anchor-based detectors. However, the anchor-based detectors typically suffer from a severe imbalance issue be-tween the positive and negative anchor boxes. To address this issue, in this work we extend the horizontal keypoint-based object detector to the oriented object detection task. In particular, we first detect the center keypoints of the objects, based on which we then regress the box boundary-aware vectors (BBAVectors) to capture the oriented bounding boxes. The box boundary-aware vectors are distributed in the four quadrants of a Cartesian coordinate system for all arbitrarily oriented objects. To relieve the difficulty of learning the vectors in the corner cases, we further classify the oriented bounding boxes into horizontal and rotational bounding boxes. In the experiment, we show that learning the box boundary-aware vectors is superior to directly predicting the width, height, and angle of an oriented bounding box, as adopted in the baseline method. Besides, the proposed method competes favorably with state-of-the-art methods. Code is available at https://github.com/yijingru/BBAVectors-Oriented-Object-Detection.
@article{
title = {Oriented object detection in aerial images with box boundary-aware vectors},
type = {article},
year = {2021},
pages = {2149-2158},
id = {cfa86472-82f1-3cbd-a61b-f7bd21e2fcf0},
created = {2024-03-27T10:37:22.089Z},
file_attached = {true},
profile_id = {bfbbf840-4c42-3914-a463-19024f50b30c},
group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},
last_modified = {2024-03-27T10:37:34.545Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {true},
hidden = {false},
folder_uuids = {df28411a-ed7f-4991-8358-d39685eb4bf0},
private_publication = {false},
abstract = {Oriented object detection in aerial images is a challenging task as the objects in aerial images are displayed in arbitrary directions and are usually densely packed. Cur-rent oriented object detection methods mainly rely on two-stage anchor-based detectors. However, the anchor-based detectors typically suffer from a severe imbalance issue be-tween the positive and negative anchor boxes. To address this issue, in this work we extend the horizontal keypoint-based object detector to the oriented object detection task. In particular, we first detect the center keypoints of the objects, based on which we then regress the box boundary-aware vectors (BBAVectors) to capture the oriented bounding boxes. The box boundary-aware vectors are distributed in the four quadrants of a Cartesian coordinate system for all arbitrarily oriented objects. To relieve the difficulty of learning the vectors in the corner cases, we further classify the oriented bounding boxes into horizontal and rotational bounding boxes. In the experiment, we show that learning the box boundary-aware vectors is superior to directly predicting the width, height, and angle of an oriented bounding box, as adopted in the baseline method. Besides, the proposed method competes favorably with state-of-the-art methods. Code is available at https://github.com/yijingru/BBAVectors-Oriented-Object-Detection.},
bibtype = {article},
author = {Yi, Jingru and Wu, Pengxiang and Liu, Bo and Huang, Qiaoying and Qu, Hui and Metaxas, Dimitris},
doi = {10.1109/WACV48630.2021.00220},
journal = {Proceedings - 2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021}
}
Downloads: 0
{"_id":"coD2pxdxstqTRbRXs","bibbaseid":"yi-wu-liu-huang-qu-metaxas-orientedobjectdetectioninaerialimageswithboxboundaryawarevectors-2021","author_short":["Yi, J.","Wu, P.","Liu, B.","Huang, Q.","Qu, H.","Metaxas, D."],"bibdata":{"title":"Oriented object detection in aerial images with box boundary-aware vectors","type":"article","year":"2021","pages":"2149-2158","id":"cfa86472-82f1-3cbd-a61b-f7bd21e2fcf0","created":"2024-03-27T10:37:22.089Z","file_attached":"true","profile_id":"bfbbf840-4c42-3914-a463-19024f50b30c","group_id":"5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1","last_modified":"2024-03-27T10:37:34.545Z","read":false,"starred":false,"authored":false,"confirmed":"true","hidden":false,"folder_uuids":"df28411a-ed7f-4991-8358-d39685eb4bf0","private_publication":false,"abstract":"Oriented object detection in aerial images is a challenging task as the objects in aerial images are displayed in arbitrary directions and are usually densely packed. Cur-rent oriented object detection methods mainly rely on two-stage anchor-based detectors. However, the anchor-based detectors typically suffer from a severe imbalance issue be-tween the positive and negative anchor boxes. To address this issue, in this work we extend the horizontal keypoint-based object detector to the oriented object detection task. In particular, we first detect the center keypoints of the objects, based on which we then regress the box boundary-aware vectors (BBAVectors) to capture the oriented bounding boxes. The box boundary-aware vectors are distributed in the four quadrants of a Cartesian coordinate system for all arbitrarily oriented objects. To relieve the difficulty of learning the vectors in the corner cases, we further classify the oriented bounding boxes into horizontal and rotational bounding boxes. In the experiment, we show that learning the box boundary-aware vectors is superior to directly predicting the width, height, and angle of an oriented bounding box, as adopted in the baseline method. Besides, the proposed method competes favorably with state-of-the-art methods. Code is available at https://github.com/yijingru/BBAVectors-Oriented-Object-Detection.","bibtype":"article","author":"Yi, Jingru and Wu, Pengxiang and Liu, Bo and Huang, Qiaoying and Qu, Hui and Metaxas, Dimitris","doi":"10.1109/WACV48630.2021.00220","journal":"Proceedings - 2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021","bibtex":"@article{\n title = {Oriented object detection in aerial images with box boundary-aware vectors},\n type = {article},\n year = {2021},\n pages = {2149-2158},\n id = {cfa86472-82f1-3cbd-a61b-f7bd21e2fcf0},\n created = {2024-03-27T10:37:22.089Z},\n file_attached = {true},\n profile_id = {bfbbf840-4c42-3914-a463-19024f50b30c},\n group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},\n last_modified = {2024-03-27T10:37:34.545Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {true},\n hidden = {false},\n folder_uuids = {df28411a-ed7f-4991-8358-d39685eb4bf0},\n private_publication = {false},\n abstract = {Oriented object detection in aerial images is a challenging task as the objects in aerial images are displayed in arbitrary directions and are usually densely packed. Cur-rent oriented object detection methods mainly rely on two-stage anchor-based detectors. However, the anchor-based detectors typically suffer from a severe imbalance issue be-tween the positive and negative anchor boxes. To address this issue, in this work we extend the horizontal keypoint-based object detector to the oriented object detection task. In particular, we first detect the center keypoints of the objects, based on which we then regress the box boundary-aware vectors (BBAVectors) to capture the oriented bounding boxes. The box boundary-aware vectors are distributed in the four quadrants of a Cartesian coordinate system for all arbitrarily oriented objects. To relieve the difficulty of learning the vectors in the corner cases, we further classify the oriented bounding boxes into horizontal and rotational bounding boxes. In the experiment, we show that learning the box boundary-aware vectors is superior to directly predicting the width, height, and angle of an oriented bounding box, as adopted in the baseline method. Besides, the proposed method competes favorably with state-of-the-art methods. Code is available at https://github.com/yijingru/BBAVectors-Oriented-Object-Detection.},\n bibtype = {article},\n author = {Yi, Jingru and Wu, Pengxiang and Liu, Bo and Huang, Qiaoying and Qu, Hui and Metaxas, Dimitris},\n doi = {10.1109/WACV48630.2021.00220},\n journal = {Proceedings - 2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021}\n}","author_short":["Yi, J.","Wu, P.","Liu, B.","Huang, Q.","Qu, H.","Metaxas, D."],"urls":{"Paper":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c/file/4de5fa4f-a1b7-79e3-007f-1c084a38079e/Yi_Oriented_Object_Detection_in_Aerial_Images_With_Box_Boundary_Aware_Vectors_WACV_2021_paper.pdf.pdf"},"biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","bibbaseid":"yi-wu-liu-huang-qu-metaxas-orientedobjectdetectioninaerialimageswithboxboundaryawarevectors-2021","role":"author","metadata":{"authorlinks":{}},"downloads":0},"bibtype":"article","biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","dataSources":["2252seNhipfTmjEBQ"],"keywords":[],"search_terms":["oriented","object","detection","aerial","images","box","boundary","aware","vectors","yi","wu","liu","huang","qu","metaxas"],"title":"Oriented object detection in aerial images with box boundary-aware vectors","year":2021}