Multiple Cues-Based Robust Visual Object Tracking Method

• DOI: 10.3390/electronics11030345
• Publication Date: 2022-01-26T01:38:58Z
• Authors: Baber Khan; Abdul Jalil; Ahmad Ali; Khaled Alkhaledi; Khizer Mehmood; Khalid Mehmood Ch...; Maria Murad; Hanan Tariq; Ahmed M. El-Sherb...
• Keywords: Robustness; BitTorrent tracker; Active appearance model; Tracking (education)

Visual object tracking is still considered a challenging task in computer vision research society. The of interest undergoes significant appearance changes because illumination variation, deformation, motion blur, background clutter, and occlusion. Kernelized correlation filter- (KCF) based schemes have shown good performance recent years. accuracy robustness these trackers can be further enhanced by incorporating multiple cues from the response map. Response map computation complementary step KCF-based schemes, it contains bundle information. majority methods on KCF estimate target location fetching single cue-like peak value This paper proposes to mine in-depth fetch about model. Furthermore, new criterion hybridization i.e., average energy (APCE) confidence squared (CSRM), presented enhance efficiency. We update following modules hybridized criterion: (i) occlusion detection, (ii) adaptive learning rate adjustment, (iii) drift handling using rate, (iv) handling, (v) scale estimation. integrate all propose scheme. proposed tracker evaluated videos selected three standard datasets, OTB-50, OTB-100, TC-128. A comparison scheme with other state-of-the-art also this paper. Our method improved considerably achieving center error 16.06, distance precision 0.889, overlap success 0.824.