Saliency prediction in the coherence theory of attention

Abstract

In the coherence theory of attention, introduced by Rensink, O'Regan, and Clark (2000), a coherence field is defined by a hierarchy of structures supporting the activities taking place across the different stages of visual attention. At the interface between low level and mid-level attention processing stages are the proto-objects; these are generated in parallel and collect features of the scene at specific location and time. These structures fade away if the region is no further attended by attention. We introduce a method to computationally model these structures. Our model is based experimentally on data collected in dynamic 3D environments via the Gaze Machine, a gaze measurement framework. This framework allows to record pupil motion at the required speed and projects the point of regard in the 3D space (Pirri, Pizzoli, & Rudi, 2011; Pizzoli, Rigato, Shabani, & Pirri, 2011). To generate proto-objects the model is extended to vibrating circular membranes whose initial displacement is generated by the features that have been selected by classification. The energy of the vibrating membranes is used to predict saliency in visual search tasks. © 2013 Elsevier B.V

Observation of environments with different restorative potential results in differences in eye patron movements and pupillary size

“…Eye tracking is a relatively novel approach in landscape architecture and environmental psychology (Nordh et al, 2013). An eye-tracking is a very useful tool to determine eye behavior (Ntouskos et al, 2013). It represents an objective, direct link between the stimuli and the participant.…”

Confidence driven TGV fusion

“…The average values reported here are arithmetic averages in order to be consistent with the values reported on the website of the KITTI benchmark. One can notice that the proposed model performs better in all the combinations apart from the combination [43] & [21]. This suggests that the proposed model is robust with respect to registration errors.…”

“…To estimate the camera motion, we considered two different stereocamera localization methods in order to recover the relative transformations between the reference and the other views. The first is based on [22], and the second is the one used in [43] for the localization of a head-mounted stereo-camera.…”

Deep Execution Monitor for Robot Assistive Tasks

“…The introduction of an attention mechanism [3,31,30] has improved sequence to sequence models essentially for neural translation and also for image captioning. Attention mechanisms for robot execution have been studied in [35], and here in particular we base our approach on the attention mechanism to exploit the task context.…”