Putting a Kinder Face on Public Cameras

Public cameras are ubiquitous in the 21st century with tens of thousands in many major cities and roadways. Currently these are used unilaterally for viewing only - there is no means for reciprocal communications. We propose that public camera networks will evolve toward supporting interactivity and hosting multiple tasks. Three tasks that can be built upon camera networks are: 1) surveillance (the current and most common task), 2) response to solicited help and information, and 3) public gaming. Video analysis that is already being performed for the surveillance task can also be used for facilitating communications required by the second and third tasks, and for measuring aspects of interaction. We describe our video analysis approach, which tracks motion flow rather than discrete moving objects. We demonstrate how motion flow yields video analytics trend and anomaly results that are used in the help and gaming tasks, and we describe results of interaction by the public.