Abstract:Visual object tracking is a complicated problem due to varied challenges such as occlusion, background clutter, target appearance changes, abrupt target motion, and illumination variations. Different tracking algorithms perform well on different challenges because of their unique strengths. In this paper, a dual-integration framework that integrates the strengths of multiple models while avoiding their weaknesses is proposed. In the proposed framework, several evaluation criteria and multiple models updated via different processes are combined. To distinguish the target from the distractors caused by different criteria, the motion dynamic model and the forward-backward analysis is introduced to provide information that is complementary to appearance information. In addition, for discriminating occlusion, a spatial-temporal occlusion-aware approach is further proposed. The detected occlusion results are applied in avoiding contamination of the appearance model. Extensive experiments on multiple benchmarks demonstrate that this proposed method improves the overlap rate of a hand-crafted feature-based tracker with relative gains of 4%, 1.6%, 1.9%, 2.2% and 8.5% on OTB-2015, OTB-2013, Temple-Color, UAV123 and UAV20L, respectively. Also, the experimental results demonstrate that our approach outperforms a deep feature-based tracker in overlap rate by 1.4%, 1.6%, 2.3%, 2.5% and 3.5% on OTB-2015, OTB-2013, Temple-Color, UAV123 and UAV20L, respectively.

Object Tracking with Adaptive Multicue Incremental Visual Tracker

Object tracking with adaptive multicue incremental visual tracker

Robust visual tracking via weighted incremental subspace learning

Weighted Incremental Subspace Learning Algorithm Suitable for Object Tracking

Multi-invariance appearance model for object tracking

Adaptive Object Tracking with Incremental Tensor Subspace Learning

MATI: Multimodal Adaptive Tracking Integrator for Robust Visual Object Tracking

Multi-features Guided Robust Visual Tracking.

Robust Visual Tracking Via Spatio-Temporal Cue Integration

Robust Visual Tracking with Incremental Subspace Learning Sparse Model

Appearance-based subspace learning model using incremental PCA in object tracking

A Joint Object Tracking Framework with Incremental and Multiple Instance Learning

Object Tracking Based on Adaptive Multi-Cue Integration Mean Shift

VisuaI Tracking Using Adaptive Structure ModeI Based on Spatio-TemporaI Context

Adaptive Incremental Principal Component Analysis Visual Tracking Method Based on Temporal Characteristics

Dual Integration of Multi‐model with Spatial‐temporal Occlusion‐awareness for Visual Object Tracking

Object Tracking Based on Incremental Bi-2DPCA Learning with Sparse Structure.

A Parallel and Robust Object Tracking Approach Synthesizing Adaptive Bayesian Learning and Improved Incremental Subspace Learning

Robust object tracking by adaptive models combination

Robust Incremental Subspace Learning for Object Tracking

Incremental Pairwise Discriminant Analysis Based Visual Tracking