Interactive Human Behavior Analysis in Open or Public Spaces

This event is now over. This page remains for historical records of this workshop.

The program for the workshop is available here.

In the past years, a lot of efforts in surveillance and open space analysis have focused on traditional computer vision problems like scene modeling or object detection and tracking. Not much work however has been done on human behavior recognition; exceptions are predefined simple activities (running, jumping, left luggage...) and single-person trajectory-based analysis. However, in open spaces, whether public or private, humans exhibit a much larger and interesting range of behaviors, from their interaction with the environment (how groups of people occupy the space or how they manipulate/use objects within it) to the way they communicate with each other. Such behaviors can be captured from multiple sensor logs processed either separately or in conjunction, such as video, audio, proximity, equipment, or other in-situ sensors, or, thanks to the pervasiveness of mobile devices, through the analysis of interaction user practices captured by enhanced location-aware sensors (e.g mobile phone or sociometric badges). These sensors along with improvements in video sensor technology in terms of resolution definitively enables improvements in the types of automated analysis that can be done. Such analysis will provide useful information to retailers, sociologists, public planning authorities, architects, as well as surveillance and security applications. A large number of projects, including EU-funded projects such as ADABTS (2009-2013), PROMETHEUS (2008-2010), VANAHEIM (2010-1013), or national projects like SWEET-HOME (2009-2012) are appearing that address these issues specifically so there is a need to foster this new research in a forum that is sympathetic to the potential applications of such research.

The goal of the workshop is to bring together experts and researchers from different fields (computer scientists, psychologists, sociologists, governmental agencies and industry) to share their experience and expertise about the opportunities on the development of tools for automated social analysis in open spaces. In particular, we would like to foster discussions between researchers in the fields of human-behavior understanding (for example, but not limited to sociologists), visual processing, multi-modal or multi-sensor processing, and potential commercial end-users in order to identify the needs, challenges and avenues of research in this field.

Sponsors

European project - FP7-248907

AnaSID: European project - PIEF-GA-2009-255609

Publisher