Real-time people counting from depth imagery of crowded environments

Abstract

In this paper we describe a system for automatic people counting in crowded environments. The approach we propose is a counting-by-detection method based on depth imagery. It is designed to be deployed as an autonomous appliance 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 automatically estimated ground plane. The system runs in realtime, at a frame-rate of about 20 fps. We collected a dataset of RGB-D sequences representing three typical and challenging 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.

Automatic surveillance in transportation hubs: No longer just about catching the bad guy

“…Approaches that use local features (Bondi, Seidenari, Bagdanov, & Del Bimbo, 2014;Kilambi, Ribnick, Joshi, Masoud, & Papanikolopoulos, 2008;Lempitsky & Zisserman, 2010;Ryan, Denman, Fookes, & Sridharan, 2009;2010;Zhang & Zhang, 2014) can overcome many of the limitations of holistic approaches. The approaches of (Bondi et al, 2014;Zhang & Zhang, 2014) use a counting by detection approach, where a learned detector for a region such as the head and shoulders is used to locate and count all people in the scene.…”

“…The approaches of (Bondi et al, 2014;Zhang & Zhang, 2014) use a counting by detection approach, where a learned detector for a region such as the head and shoulders is used to locate and count all people in the scene. While this works well for uncluttered environments, it does not perform well in dense crowds.…”

Communication and computation inter-effects in people counting using intelligence partitioning

“…Alternative methods such as analysis of WiFi channel use [10], or a combination of infrared lasers alongside an IR camera to detect the IR rays displacement [11] have been implemented, providing an approximate count, but without distinguishing between people and other moving elements that can be present in the region of interest. Instead, image-based people counting models introduced in [12][13][14][15] rely on depth imagery to overcome occlusion limitations, while using GPU as processing element. The emphasis of their approach is on the accuracy of detection; hence, they propose models that are computationally intensive in devices that cannot be battery operated.…”

Utilizing Attributed Graph Representation in Object Detection and Tracking for Indoor Range Sensor Surveillance Cameras

“…In the proposed method by Jafari et al [7] the depth information is utilized via an RGB-D camera mounted on the person's head. Fu et al [8] and Bondi et al [9] use depth information in a people counting system to estimate the position of head and shoulders of the detected objects to accurately locate the human objects to be counted.…”