Abstract:
Crowds of people offer multitudes of stimuli for attracting gaze and attention.
A crowded scene is considered to be formed with a small or a large number
of objects or people standing together in a scene. What influences our attention in crowds can be different in static and dynamic conditions as the
visual attention system of the humans is automatically directed to focus on
the most pertinent stimuli in being presented in the scene. Many researches
have concluded the preferences of people which viewing at the crowds more
in the form of static images and a few in videos. We believe that the crowds
are supposed to be dynamic and not static. Hence we constructed the dataset
of real life crowd videos and transformed them into MTV-style video clips
by combining the clippets together using jump cuts. We chose the participants with no specific knowledge and recorded their eye movements. It is
usually accepted that every individuals attention is instantly influenced by
the faces of the people. We are interested in exploring if this statement holds
in conflation with crowd density (the number of people per square area). To
this end, we would like to understand this relationship between human gaze
and varying crowd densities using the fixation data from eye tracker recordings. The statistical fixation analysis will provide insights into the factors
influencing gaze in varying levels of crowds.
