Real-time people counting from depth imagery of crowded environments

In this paper we describe a system for automatic people counting in crowded environments. The approach we pro-pose is a counting-by-detection method based on depth im-agery. It is designed to be deployed as an autonomous ap-pliance for crowd analysis in video surveillance application scenarios. Our system performs foreground/background segmentation on depth image streams in order to coarsely segment persons, then depth information is used to localize head candidates which are then tracked in time on an auto-matically estimated ground plane. The system runs in real-time, at a frame-rate of about 20 fps. We collected a dataset of RGB-D sequences representing three typical and chal-lenging surveillance scenarios, including crowds, queuing and groups. An extensive comparative evaluation is given between our system and more complex, Latent SVM-based head localization for person counting applications. 1