Robust single target tracking using determinantal point process observations

• DOI: 10.21307/ijssis-2020-001
• Publication Date: 2020-02-01T09:00:17Z
• Authors: S. Hernández; P. Sallis
• Keywords: Discriminative model; Robustness; Determinantal point process; Tracking (education)

Abstract The efficiency and robustness of modern visual tracking systems are largely dependent on the object detection system at hand. Bernoulli Multi-Bernoulli filters have been proposed for without explicit detections (image observations). However, these previous approaches do not fully exploit discriminative features tracking. In this paper, we propose a novel filter with determinantal point processes observations. observation model can select groups high scores low correlation among observed features; thus achieving robust filter.