Output details
11 - Computer Science and Informatics
University of Edinburgh
A computer vision model for visual-object-based attention and eye movements
<23> Originality: Previous computational models were based on location-based computation of salience (interestingness). This paper is about how the first computational model of object-based salience can be used to drive saccades (gaze shifts).
Significance: This research is the first computational model based on an major alternative psychological attention theory, and produces better visual selection than the space based theory.
Rigour: A full computational model is given covering both mathematical and algorithmic aspects, suitable for re-implementation. It was evaluated both on commonly-used images and new images (about 10). Evaluation showed that saccades were more likely to land on interesting objects.