COBRA (Crowd Behavior Analyzer) is a people monitoring and profiling system for crowded environments. It is meant to work from multiple cameras “in the wild”, in uncontrolled capturing conditions. To track users and build a continuous flows across the cameras, the system has to re-identify the subjects while they appear in the different views. This will allow to perform mobility and behaviour analysis of groups of individuals.
The subjects’ re-identification is performed using face recognition technologies; indeed, faces represent a more discriminative and robust clue with respect to the appearance of the body to recognize and track moving subjects, mostly in crowded environments in which people can have the same clothing. General demographic information will be also collected so as to enrich the gathered knowledge and perform soft-biometrics.
With respect to the current technological state of the art for crowd analysis, the COBRA system is scalable to multiple cameras, relying on the robustness of facial descriptors. Grounding on the informations provided by the system, it will become easier to manage and organize spaces and environments, and ultimately obtain an automated monitoring.
The COBRA system has a wide variety of possible applications, ranging from security and surveillance to the spatial distribution of goods, and it is scalable to smaller or larger environments.