Real-Time Compressive Tracking

Abstract

Abstract. It is a challenging task to develop effective and efficient appearance models for robust object tracking due to factors such as pose variation, illumination change, occlusion, and motion blur. Existing online tracking algorithms often update models with samples from observations in recent frames. While much success has been demonstrated, numerous issues remain to be addressed. First, while these adaptive appearance models are data-dependent, there does not exist sufficient amount of data for online algorithms to learn at the outset. Second, online tracking algorithms often encounter the drift problems. As a result of self-taught learning, these mis-aligned samples are likely to be added and degrade the appearance models. In this paper, we propose a simple yet effective and efficient tracking algorithm with an appearance model based on features extracted from the multi-scale image feature space with data-independent basis. Our appearance model employs nonadaptive random projections that preserve the structure of the image feature space of objects. A very sparse measurement matrix is adopted to efficiently extract the features for the appearance model. We compress samples of foreground targets and the background using the same sparse measurement matrix. The tracking task is formulated as a binary classification via a naive Bayes classifier with online update in the compressed domain. The proposed compressive tracking algorithm runs in real-time and performs favorably against state-of-the-art algorithms on challenging sequences in terms of efficiency, accuracy and robustness.

Consistent multi-layer subtask tracker via hyper-graph regularization

“…The Struck [8] ranks top in the recent benchmark [14] , and it learns a kernelized structured output support vector machine online. Recently, an efficient tracking algorithm [7] based on compressive sensing theory is proposed. Correlation filter has been widely used in object tracking area with good performance and high speed [38][39][40][41][42] .…”

“…It is an important and fundamental topic in computer vision, and has many applications including intelligent surveillance, motion recognition, robotic, human computer interface (HCI), augmented reality (AR), etc. Although great processes have been made and many tracking algorithms have been proposed [1][2][3][4][5][6][7][8] , it remains an open problem to design a robust tracker in the real-world scenarios due to severe occlusions, large appearance changes, illumination changes, background clutter and abrupt motion, etc.…”

Low-Rank Representation with Graph Constraints for Robust Visual Tracking

“…Incremental Visual Tracking (IVT) [6], Consistent Low-Rank Sparse Tracker (CLRST) [27], Multi-task Sparse Tracking (MTT) [26], Superpixel tracking (ST) [33], Kernelized Correlation Filters (KCF) [34] and Compressive Tracking (CT) [35]. IVT applies PCA to achieve and update the subspace online.…”

Self-taught learning of a deep invariant representation for visual tracking via temporal slowness principle

“…The competing trackers are Adaptive Structural Local-sparse Appearance (ASLA) tracker [37], Compressive Tracker (CT) [46], Deep Learning Tracker (DLT) [20], Incremental Visual Tracker (IVT) [3], Online Discriminative Feature Selection (ODFS) tracker [47], Partial Least Squares (PLS) tracker [48], Sparse Prototypes Tracker (SPT) [49], and Tracking-Learning-Detection (TLD) [50]. We run the experiments based on the codes provided by the authors.…”