PROJECT TITLE :

Gracker: A Graph-based Planar Object Tracker - 2017

ABSTRACT:

Matching-based mostly algorithms have been commonly employed in planar object tracking. They typically model a planar object as a collection of keypoints, and then realize correspondences between keypoint sets via descriptor matching. In previous work, unary constraints on appearances or locations are typically used to guide the matching. But, these approaches rarely utilize structure data of the item, and are thus littered with various perturbation factors. In this paper, we have a tendency to proposed a graph-based mostly tracker, named Gracker, that is in a position to fully explore the structure info of the thing to enhance tracking performance. We model a planar object as a graph, rather than a straightforward collection of keypoints, to represent its structure. Then, we have a tendency to reformulate tracking as a sequential graph matching process, which establishes keypoint correspondence in a geometric graph matching manner. For analysis, we have a tendency to compare the proposed Gracker with state-of-the-art planar object trackers on three benchmark datasets: two public ones and a newly collected one. Experimental results show that Gracker achieves strong tracking results against various environmental variations, and outperforms other algorithms generally on the datasets.