Online Deformable Object Tracking Based on Structure – Aware Hyper – Graph
Abstract of Online Deformable Object Tracking Based on Structure – Aware Hyper – Graph
Online Deformable Object Tracking Based on Structure – Aware Hyper – graph,Recent advances in online visual tracking focus on designing part-based model to handle the deformation and occlusion challenges.
This paper describes a new and efficient method for online deformable object tracking.
The experimental result of the proposed method shows considerable improvement in performance over the state-of-the-art tracking methods.
Online visual tracking is an important step toward fully automatic understanding of videos, which finds wide applications in video surveillance, behavior analysis, human computer interaction, to name a few.
In spite of significant progress in the recent years, online tracking a deformable object accurately remains a difficult problem.
Currently, the predominant approaches in online visual tracking aim to obtain a bounding box of the tracking target.
Many methods are built on models which focus on capturing appearance variation of the tracking target in the bounding box, e.g., correlation filters, subspace learning, online boostingand sparse representation.
In spite of significant progress in the recent years, online tracking a deformable object accurately remains a difficult problem.
Currently, the predominant approaches in online visual tracking aim to obtain a bounding box of the tracking target.
Conclusion
In this paper, we describe a structure-aware hyper-graph based tracker. There are a few directions we would like to further extend the current work.
First, in the current method, we only consider temporal higher-order dependencies among the parts.
As a next step, we will also investigate incorporating spatial higher-order dependencies among the parts.
Considering both spatial and temporal dependencies among parts is expected to further improve tracking performance and robustness under deformation and occlusion.
