Please use this identifier to cite or link to this item: http://localhost/handle/Hannan/220291
Title: Geometric Hypergraph Learning for Visual Tracking
Authors: Dawei Du;Honggang Qi;Longyin Wen;Qi Tian;Qingming Huang;Siwei Lyu
Year: 2017
Publisher: IEEE
Abstract: Graph-based representation is widely used in visual tracking field by finding correct correspondences between target parts in different frames. However, most graph-based trackers consider pairwise geometric relations between local parts. They do not make full use of the target's intrinsic structure, thereby making the representation easily disturbed by errors in pairwise affinities when large deformation or occlusion occurs. In this paper, we propose a geometric hypergraph learning-based tracking method, which fully exploits high-order geometric relations among multiple correspondences of parts in different frames. Then visual tracking is formulated as the mode-seeking problem on the hypergraph in which vertices represent correspondence hypotheses and hyperedges describe high-order geometric relations among correspondences. Besides, a confidence-aware sampling method is developed to select representative vertices and hyperedges to construct the geometric hypergraph for more robustness and scalability. The experiments are carried out on three challenging datasets (VOT2014, OTB100, and Deform-SOT) to demonstrate that our method performs favorably against other existing trackers.
URI: http://localhost/handle/Hannan/220291
volume: 47
issue: 12
More Information: 4182,
4195
Appears in Collections:2017

Files in This Item:
File SizeFormat 
7748448.pdf2.96 MBAdobe PDF
Title: Geometric Hypergraph Learning for Visual Tracking
Authors: Dawei Du;Honggang Qi;Longyin Wen;Qi Tian;Qingming Huang;Siwei Lyu
Year: 2017
Publisher: IEEE
Abstract: Graph-based representation is widely used in visual tracking field by finding correct correspondences between target parts in different frames. However, most graph-based trackers consider pairwise geometric relations between local parts. They do not make full use of the target's intrinsic structure, thereby making the representation easily disturbed by errors in pairwise affinities when large deformation or occlusion occurs. In this paper, we propose a geometric hypergraph learning-based tracking method, which fully exploits high-order geometric relations among multiple correspondences of parts in different frames. Then visual tracking is formulated as the mode-seeking problem on the hypergraph in which vertices represent correspondence hypotheses and hyperedges describe high-order geometric relations among correspondences. Besides, a confidence-aware sampling method is developed to select representative vertices and hyperedges to construct the geometric hypergraph for more robustness and scalability. The experiments are carried out on three challenging datasets (VOT2014, OTB100, and Deform-SOT) to demonstrate that our method performs favorably against other existing trackers.
URI: http://localhost/handle/Hannan/220291
volume: 47
issue: 12
More Information: 4182,
4195
Appears in Collections:2017

Files in This Item:
File SizeFormat 
7748448.pdf2.96 MBAdobe PDF
Title: Geometric Hypergraph Learning for Visual Tracking
Authors: Dawei Du;Honggang Qi;Longyin Wen;Qi Tian;Qingming Huang;Siwei Lyu
Year: 2017
Publisher: IEEE
Abstract: Graph-based representation is widely used in visual tracking field by finding correct correspondences between target parts in different frames. However, most graph-based trackers consider pairwise geometric relations between local parts. They do not make full use of the target's intrinsic structure, thereby making the representation easily disturbed by errors in pairwise affinities when large deformation or occlusion occurs. In this paper, we propose a geometric hypergraph learning-based tracking method, which fully exploits high-order geometric relations among multiple correspondences of parts in different frames. Then visual tracking is formulated as the mode-seeking problem on the hypergraph in which vertices represent correspondence hypotheses and hyperedges describe high-order geometric relations among correspondences. Besides, a confidence-aware sampling method is developed to select representative vertices and hyperedges to construct the geometric hypergraph for more robustness and scalability. The experiments are carried out on three challenging datasets (VOT2014, OTB100, and Deform-SOT) to demonstrate that our method performs favorably against other existing trackers.
URI: http://localhost/handle/Hannan/220291
volume: 47
issue: 12
More Information: 4182,
4195
Appears in Collections:2017

Files in This Item:
File SizeFormat 
7748448.pdf2.96 MBAdobe PDF