Pooling pedestrian tracking information for a high-level scene analysis

Abstract

Capturing the dynamics of pedestrians, vehicles, and other objects on a video sequence, leads to further knowledge of what is happening on a scene. In this paper, a targlet framework is developed to detect pedestrians, which associates each person with an autonomous entity called targlet (target+aglet), and tracks those targlets through the sequence. The particularity of our methodology is the use of a movement feature space (MFS) to generate descriptors for the classifier and the tracker.