Centroid Based Human Annotation for Object Tracking
Surabhi Sukumar, T M Libish · 2018 2nd International Conference on Trends in Electronics and Informatics (ICOEI) · 2018
Visual object tracking is most important in surveillance systems. Trackers should be robust to occlusion while tracking a particular object. There are number of approaches to track objects. This paper introduces a new method that use centroid based human annotation to track moving objects in unseen videos with the help of motion patterns. It does not require object detectors and manual initialization. Centroid based method can handle up to 90% of occlusion. In this paper a library of annotated videos are created and a query video is tested against the library. Matching is done using Greedy algorithm.