The second paper: MCPF Multi-task Correlation Particle Filter for
Robust Object Tracking/ Author: Tianzhu Zhang, Changsheng Xu,
Ming-Hsuan Yang/ Publication information: CVPR2017
Outline: MCPF utilizes the advantage particle filters in VOT. Particle
filters can maintain good information in the previous frame, the
probability distribution and the scale information. Multiple particle filters
in different features will have the same moving trend between the
continuous frame. Consequently, multi-task correlation filter helps reduce
the number of particles and cut down the computation burden. And the
correlation filters from circular samples can also shepherd the whole
particles.
Advantages: The experiments prove that the satisfactory accuracy and
robustness of MCPF. MCPF performs very good when encountered with
many challenges.
Disadvantages: The speed is faster than C-COT, but still less than 1 fps.
Robust Object Tracking/ Author: Tianzhu Zhang, Changsheng Xu,
Ming-Hsuan Yang/ Publication information: CVPR2017
Outline: MCPF utilizes the advantage particle filters in VOT. Particle
filters can maintain good information in the previous frame, the
probability distribution and the scale information. Multiple particle filters
in different features will have the same moving trend between the
continuous frame. Consequently, multi-task correlation filter helps reduce
the number of particles and cut down the computation burden. And the
correlation filters from circular samples can also shepherd the whole
particles.
Advantages: The experiments prove that the satisfactory accuracy and
robustness of MCPF. MCPF performs very good when encountered with
many challenges.
Disadvantages: The speed is faster than C-COT, but still less than 1 fps.