TY - JOUR
T1 - Effective visual tracking by pairwise metric learning
AU - Deng, Chenwei
AU - Wang, Baoxian
AU - Lin, Weisi
AU - Huang, Guang Bin
AU - Zhao, Baojun
N1 - Publisher Copyright:
© 2017 Elsevier B.V.
PY - 2017/10/25
Y1 - 2017/10/25
N2 - For robust visual tracking, appearance modeling should be able to well separate the object from its backgrounds, while accurately adapt to its appearance variations. However, most of the existing tracking methods mainly focus on one of the two aspects; or design two different modules to combine them with the price of double computational cost. In this paper, by using pairwise metric learning, we present a novel appearance model for robust visual tracking. Specifically, visual tracking is viewed as a pairwise regression problem, and extreme learning machine (ELM) is utilized to construct the pairwise regression framework. In ELM-based pairwise training, two constraints are enforced: the target observations must have different regression outputs from those background ones; while the various target observations during tracking should have approximate regression outputs. Thus, the discriminative and generative capabilities are fully considered in a single object tracking model. Moreover, online sequential ELM (OS-ELM) is used to update the resulting appearance model, thereby leading to a more robust tracking process. Extensive experimental evaluations on challenging video sequences demonstrate the effectiveness and efficiency of the proposed tracker.
AB - For robust visual tracking, appearance modeling should be able to well separate the object from its backgrounds, while accurately adapt to its appearance variations. However, most of the existing tracking methods mainly focus on one of the two aspects; or design two different modules to combine them with the price of double computational cost. In this paper, by using pairwise metric learning, we present a novel appearance model for robust visual tracking. Specifically, visual tracking is viewed as a pairwise regression problem, and extreme learning machine (ELM) is utilized to construct the pairwise regression framework. In ELM-based pairwise training, two constraints are enforced: the target observations must have different regression outputs from those background ones; while the various target observations during tracking should have approximate regression outputs. Thus, the discriminative and generative capabilities are fully considered in a single object tracking model. Moreover, online sequential ELM (OS-ELM) is used to update the resulting appearance model, thereby leading to a more robust tracking process. Extensive experimental evaluations on challenging video sequences demonstrate the effectiveness and efficiency of the proposed tracker.
KW - Appearance modeling
KW - Extreme learning machine
KW - Online sequential updating
KW - Pairwise metric learning
KW - Robust visual tracking
UR - http://www.scopus.com/inward/record.url?scp=85013436177&partnerID=8YFLogxK
U2 - 10.1016/j.neucom.2016.05.115
DO - 10.1016/j.neucom.2016.05.115
M3 - Article
AN - SCOPUS:85013436177
SN - 0925-2312
VL - 261
SP - 266
EP - 275
JO - Neurocomputing
JF - Neurocomputing
ER -