Crowd density analysis with marked point processes

The abundance of video footage from surveillance systems in public spaces has become a driving force for advances in crowd image analysis. Of particular interest is crowd density analysis, where the goal is to detect and count people in a crowded scene. This is a challenging problem for a human observer when large numbers of constantly moving individuals are present. It is therefore desirable to have computational assets that can assist security personnel for real-time crowd monitoring. Automated analysis holds the potential to increase situational awareness for crowd control and public safety by providing real-time estimates of the number of people entering or exiting a venue.