Beyond Bounding-Box: Convex-hull Feature Adaptation for Oriented and Densely Packed Object Detection

被引:146
作者
Guo, Zonghao [1 ]
Liu, Chang [1 ]
Zhang, Xiaosong [1 ]
Jiao, Jianbin [1 ]
Ji, Xiangyang [2 ]
Ye, Qixiang [1 ]
机构
[1] Univ Chinese Acad Sci, Beijing, Peoples R China
[2] Tsinghua Univ, Beijing, Peoples R China
来源
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 | 2021年
关键词
ROTATION-INVARIANT; SCALE;
D O I
10.1109/CVPR46437.2021.00868
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Detecting oriented and densely packed objects remains challenging for spatial feature aliasing caused by the intersection of reception fields between objects. In this paper, we propose a convex-hull feature adaptation (CFA) approach for configuring convolutional features in accordance with oriented and densely packed object layouts. CFA is rooted in convex-hull feature representation, which defines a set of dynamically predicted feature points guided by the convex intersection over union (CIoU) to bound the extent of objects. CFA pursues optimal feature assignment by constructing convex-hull sets and dynamically splitting positive or negative convex-hulls. By simultaneously considering overlapping convex-hulls and objects and penalizing convex-hulls shared by multiple objects, CFA alleviates spatial feature aliasing towards optimal feature adaptation. Experiments on DOTA and SKU110K-R datasets show that CFA significantly outperforms the baseline approach, achieving new state-of-the-art detection performance.
引用
收藏
页码:8788 / 8797
页数:10
相关论文
共 44 条
[1]  
[Anonymous], 2015, NEURALIPS
[2]   Deformable Convolutional Networks [J].
Dai, Jifeng ;
Qi, Haozhi ;
Xiong, Yuwen ;
Li, Yi ;
Zhang, Guodong ;
Hu, Han ;
Wei, Yichen .
2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2017, :764-773
[3]  
Deng J, 2009, PROC CVPR IEEE, P248, DOI 10.1109/CVPRW.2009.5206848
[4]   Learning RoI Transformer for Oriented Object Detection in Aerial Images [J].
Ding, Jian ;
Xue, Nan ;
Long, Yang ;
Xia, Gui-Song ;
Lu, Qikai .
2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, :2844-2853
[5]   The Pascal Visual Object Classes (VOC) Challenge [J].
Everingham, Mark ;
Van Gool, Luc ;
Williams, Christopher K. I. ;
Winn, John ;
Zisserman, Andrew .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 2010, 88 (02) :303-338
[6]  
Fu C.Y., 2017, ARXIV
[7]   Precise Detection in Densely Packed Scenes [J].
Goldman, Eran ;
Herzig, Roei ;
Eisenschtat, Aviv ;
Goldberger, Jacob ;
Hassner, Tal .
2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, :5222-5231
[8]   Rotation-invariant and scale-invariant Gabor features for texture image retrieval [J].
Han, Ju ;
Ma, Kai-Kuang .
IMAGE AND VISION COMPUTING, 2007, 25 (09) :1474-1481
[9]  
Jaderberg M, 2015, ADV NEUR IN, V28
[10]  
Jarvis R. A., 1973, Information Processing Letters, V2, P18, DOI 10.1016/0020-0190(73)90020-3