Abstract
Object and human tracking in streaming videos are one of the most challenging problems in vision computing. In this article, we review some relevant machine learning algorithms and techniques for human identification and tracking in videos. We provide details on metrics and methods used in the computer vision literature for monitoring and propose a state-space representation of the object tracking problem. A proof of concept implementation of the state-space based object tracking using particle filters is presented as well. The proposed approach enables tracking objects/humans in a video, including foreground/background separation for object movement detection.
