Observing the Observers: Social Context Analysis Using Computer Vision
Colloquium in Mathematics and Computer Science
Observing the Observers: Social Context Analysis Using Computer Vision
Observing the Observers: Social Context Analysis Using Computer Vision
Meir Cohen, Technion
Sunday, December 28 |15:00 | Science Building 8, Room 424
Abstract:
It is quite common that multiple human observers visually attend to a single point of interest. This social phenomena is known as mutual awareness. Our work studies the underlying geometric constraints of mutual awareness and its related dynamics. Those constraints are used in a method that detects and tracks mutual awareness along with its related attributes. The input to the method are the gaze directions of the observers, which might be obtained by applying existing face detection and head pose estimation algorithms on images of the observers.
The suggested method is unsupervised and can deal with the general case of an uncalibrated camera in a general environment and an unconstrained activity in the scene. This is in contrast to other work on similar problems that inherently assume a known environment or a calibrated camera or a restricted occurrence in the scene.
In addition, our work attaches a social semantics to the detected mutual awareness. A deeper social interpretation is suggested by exploiting and analyzing the spatiotemporal correlations between the mutual awareness and the activity in the scene. The statistics of those social interpretation are aggregated over a long time yielding social characteristics of an individual observer, the entire group, and of the activity in the scene.
It is quite common that multiple human observers visually attend to a single point of interest. This social phenomena is known as mutual awareness. Our work studies the underlying geometric constraints of mutual awareness and its related dynamics. Those constraints are used in a method that detects and tracks mutual awareness along with its related attributes. The input to the method are the gaze directions of the observers, which might be obtained by applying existing face detection and head pose estimation algorithms on images of the observers.
The suggested method is unsupervised and can deal with the general case of an uncalibrated camera in a general environment and an unconstrained activity in the scene. This is in contrast to other work on similar problems that inherently assume a known environment or a calibrated camera or a restricted occurrence in the scene.
In addition, our work attaches a social semantics to the detected mutual awareness. A deeper social interpretation is suggested by exploiting and analyzing the spatiotemporal correlations between the mutual awareness and the activity in the scene. The statistics of those social interpretation are aggregated over a long time yielding social characteristics of an individual observer, the entire group, and of the activity in the scene.
