Depth in the visual attention modelling from the egocentric perspective of view

Miroslav Laco

Supervisor(s): Wanda Benesova

Slovak Technical University

Abstract: A research goal of computer vision scientists in the field of human visual perception is to determine and predict visual attention by the creation of various models of visual attention. An extensive research has been held in the field of visual attention modelling throughout the past years, aiming to create a complex visual attention model as close to ground-truth as possible. In this paper, we mention the benefits of user studies on the human visual attention in real world environments from the egocentric perspective of view and we introduce a novel and complete method proposal for such user studies in a laboratory. We make use of specific hardware equipment (eye-tracking glasses, Kinect and LCD projectors) and introduce our own algorithms and procedures based on computer vision theory in our method proposal. The implementation of a novel method for user studies resulted in a small novel dataset created during the first experiments. One of the benefits of the proposed novel method is the possibility to study aspects affecting visual attention that were not possible to study before. Based on the previous research in the field, we decided to conduct a research of depth influence (distance between the observer and the observed object) on human visual attention using created dataset. We claim that the depth of the scene plays significant role as an aspect of human visual attention in real world environments.
Keywords: Computer Vision, Image Processing
Full text:
Year: 2018