Fast visual tracking via dense spatio-temporal context learning

Kaihua Zhang, Lei Zhang, Qingshan Liu, David Zhang, Ming Hsuan Yang

Research output: Contribution to journalConference article

391 Citations (Scopus)

Abstract

In this paper, we present a simple yet fast and robust algorithm which exploits the dense spatio-temporal context for visual tracking. Our approach formulates the spatio-temporal relationships between the object of interest and its locally dense contexts in a Bayesian framework, which models the statistical correlation between the simple low-level features (i.e., image intensity and position) from the target and its surrounding regions. The tracking problem is then posed by computing a confidence map which takes into account the prior information of the target location and thereby alleviates target location ambiguity effectively. We further propose a novel explicit scale adaptation scheme, which is able to deal with target scale variations efficiently and effectively. The Fast Fourier Transform (FFT) is adopted for fast learning and detection in this work, which only needs 4 FFT operations. Implemented in MATLAB without code optimization, the proposed tracker runs at 350 frames per second on an i7 machine. Extensive experimental results show that the proposed algorithm performs favorably against state-of-the-art methods in terms of efficiency, accuracy and robustness.

Original languageEnglish
Pages (from-to)127-141
Number of pages15
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8693 LNCS
Issue numberPART 5
DOIs
Publication statusPublished - 2014 Jan 1
Event13th European Conference on Computer Vision, ECCV 2014 - Zurich, Switzerland
Duration: 2014 Sep 62014 Sep 12

Fingerprint

Visual Tracking
Fast Fourier transforms
Target
Fast Fourier transform
MATLAB
Robust Algorithm
Prior Information
Fast Algorithm
Confidence
Robustness
Learning
Context
Optimization
Computing
Experimental Results

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

@article{685b1cf3235747918f17259d7e815045,
title = "Fast visual tracking via dense spatio-temporal context learning",
abstract = "In this paper, we present a simple yet fast and robust algorithm which exploits the dense spatio-temporal context for visual tracking. Our approach formulates the spatio-temporal relationships between the object of interest and its locally dense contexts in a Bayesian framework, which models the statistical correlation between the simple low-level features (i.e., image intensity and position) from the target and its surrounding regions. The tracking problem is then posed by computing a confidence map which takes into account the prior information of the target location and thereby alleviates target location ambiguity effectively. We further propose a novel explicit scale adaptation scheme, which is able to deal with target scale variations efficiently and effectively. The Fast Fourier Transform (FFT) is adopted for fast learning and detection in this work, which only needs 4 FFT operations. Implemented in MATLAB without code optimization, the proposed tracker runs at 350 frames per second on an i7 machine. Extensive experimental results show that the proposed algorithm performs favorably against state-of-the-art methods in terms of efficiency, accuracy and robustness.",
author = "Kaihua Zhang and Lei Zhang and Qingshan Liu and David Zhang and Yang, {Ming Hsuan}",
year = "2014",
month = "1",
day = "1",
doi = "10.1007/978-3-319-10602-1_9",
language = "English",
volume = "8693 LNCS",
pages = "127--141",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Springer Verlag",
number = "PART 5",

}

Fast visual tracking via dense spatio-temporal context learning. / Zhang, Kaihua; Zhang, Lei; Liu, Qingshan; Zhang, David; Yang, Ming Hsuan.

In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 8693 LNCS, No. PART 5, 01.01.2014, p. 127-141.

Research output: Contribution to journalConference article

TY - JOUR

T1 - Fast visual tracking via dense spatio-temporal context learning

AU - Zhang, Kaihua

AU - Zhang, Lei

AU - Liu, Qingshan

AU - Zhang, David

AU - Yang, Ming Hsuan

PY - 2014/1/1

Y1 - 2014/1/1

N2 - In this paper, we present a simple yet fast and robust algorithm which exploits the dense spatio-temporal context for visual tracking. Our approach formulates the spatio-temporal relationships between the object of interest and its locally dense contexts in a Bayesian framework, which models the statistical correlation between the simple low-level features (i.e., image intensity and position) from the target and its surrounding regions. The tracking problem is then posed by computing a confidence map which takes into account the prior information of the target location and thereby alleviates target location ambiguity effectively. We further propose a novel explicit scale adaptation scheme, which is able to deal with target scale variations efficiently and effectively. The Fast Fourier Transform (FFT) is adopted for fast learning and detection in this work, which only needs 4 FFT operations. Implemented in MATLAB without code optimization, the proposed tracker runs at 350 frames per second on an i7 machine. Extensive experimental results show that the proposed algorithm performs favorably against state-of-the-art methods in terms of efficiency, accuracy and robustness.

AB - In this paper, we present a simple yet fast and robust algorithm which exploits the dense spatio-temporal context for visual tracking. Our approach formulates the spatio-temporal relationships between the object of interest and its locally dense contexts in a Bayesian framework, which models the statistical correlation between the simple low-level features (i.e., image intensity and position) from the target and its surrounding regions. The tracking problem is then posed by computing a confidence map which takes into account the prior information of the target location and thereby alleviates target location ambiguity effectively. We further propose a novel explicit scale adaptation scheme, which is able to deal with target scale variations efficiently and effectively. The Fast Fourier Transform (FFT) is adopted for fast learning and detection in this work, which only needs 4 FFT operations. Implemented in MATLAB without code optimization, the proposed tracker runs at 350 frames per second on an i7 machine. Extensive experimental results show that the proposed algorithm performs favorably against state-of-the-art methods in terms of efficiency, accuracy and robustness.

UR - http://www.scopus.com/inward/record.url?scp=84906482212&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84906482212&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-10602-1_9

DO - 10.1007/978-3-319-10602-1_9

M3 - Conference article

AN - SCOPUS:84906482212

VL - 8693 LNCS

SP - 127

EP - 141

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

IS - PART 5

ER -