Collaborative Correlation Tracking

Abstract

Motivation. Correlation filter based tracking has attracted many researchers' attention in recent years for high efficiency and robustness. Most existing works [1, 2, 4] focus on exploiting different characteristics with correlation filters for visual tracking, e.g., circulant structure, kernel trick, effective feature representation and context information. Despite its good performance, most of these correlation methods have two main limitations, the first is how to adjust the object scale efficiently. In order to consistently track the object, Danelljan et.al. [2] proposed a separate 1-dimensional correlation filter to estimate the target scale, but they only use the original feature space as the object representation. The second limitation is how to handle the model drift problem caused by the long-term occlusion or out-of-view, which is a very important problem for online tracking [5]. One common mechanism is to introduce a detection module which can select some effective candidates to rectify the base correlation tracker. Contributions. In this paper, we design a novel online CUR filter for detection. CUR matrix approximation computes the low rank approximation of a given matrix by using the actual rows and columns of the matrix and have been studied in the area of theoretical computer science for large matrix approximation [3]. In the long-term tracking process, all of the historical object representations can form a large data matrix for the current frame which fits for the CUR theory. The large data matrix can be fast approximated by online CUR for representing the intrinsic object structure. In this work we develop an online CUR for learning an online detection filter by random sampling. The online CUR filter can not only exploit the low rank property of object representation [7] in the spatial-temporal domain of tracking, but also project the historical object representation matrix into a subspace with error upper bound so as to achieve a robust object representation. The low rank property of object representation is prevalent in long-term tracking and could be used to alleviate the model drift. Meanwhile, we propose multi-scale kernelized correlation filter as our tracking filter by embedding the scale variation into the kernelized correlation filter while forming a separate pyramid of object representation. The collaboration between CUR filter and multi-scale kernelized correlation filter constructs the proposed tracker-Collaborative Correlation Tracker (CCT) 1 .

Vision-Based Real-Time Aerial Object Localization and Tracking for UAV Sensing System

“…The average speed of our tracker is 141 fps, which is at the same level as the fastest tracker KCF [12], however, KCF adopts a fixed tracking box, which could not reflect the scale changes of the target object during tracking. On average, our method is more than ten times faster than CT [42] and SAMF [20], five times faster than DSST [6] and CCT [44] and about two times faster than STC [41] and CN [7].…”

“…Henriques et al [12] proposed a tracker using kernelized correlation filters (KCF). Zhu et al [44] extended the KCF to a multi-scale kernelized tracker in order to deal with the scale variation. Zhang et al [41] proposed a tracker via dense spatio-temporal context learning.…”

A comparison study on the adaptive scale estimation of correlation filter-based visual tracking methods

“…[27] to reduce the computational cost, while tracking the performance is preserved. A collaborative correlation tracker is proposed in [28]. The experiment results [14,24] show that by combining the scale estimation with translation filter, their approaches outperform 19 state-of-the-art trackers in the OTB dataset.…”

The Thermal Infrared Visual Object Tracking VOT-TIR2016 Challenge Results

“…Guibo Zhu, Jinqiao Wang, Hanqing Lu {gbzhu, jqwang, luhq}@nlpr.ia.ac.cn CCFP tracker is mainly based on the idea of collaborative correlation tracking [47]. Some confident candidate proposals are generated through online detection or background modeling, and used to improve the overall tracking capability of the correlation filter-based tracker.…”