Location-Aware and Regularization-Adaptive Correlation Filters for Robust Visual Tracking.

Area of Science:

• Computer Vision
• Machine Learning

Background:

• Correlation Filters (CF) are prevalent in visual tracking.
• Current CF models often address search window estimation and filter learning separately using heuristic methods.
• This separation can compromise tracking performance due to fixed regularization and search window strategies.

Purpose of the Study:

• To propose a unified framework, Location-Aware and Regularization-Adaptive CF (LRCF), for robust visual tracking.
• To simultaneously optimize object location estimation and filter learning.
• To enhance the flexibility and effectiveness of CF-based trackers.

Main Methods:

• Developed a novel bilevel optimization model for simultaneous location estimation and filter training.
• Proved global convergence properties of the proposed bilevel formulation.
• Designed two trackers, LRCF-S and LRCF-SA, based on the LRCF framework.

Main Results:

• The LRCF framework enables collaborative optimization of object location and CF learning.
• LRCF-S and LRCF-SA demonstrate flexibility and effectiveness.
• Extensive experiments show favorable performance against state-of-the-art methods on challenging datasets.

Conclusions:

• The proposed LRCF framework offers a robust and adaptive approach to visual tracking.
• Simultaneous optimization of location and filter learning leads to improved tracking accuracy.
• LRCF trackers provide a competitive alternative to existing methods in challenging scenarios.