Saliency map augmentation with facial detection

Abstract

Visual attention is very important in human visual perception. It is the ability of a vision system to detect salient objects in an observed scene. This scientific discipline has been studied for over a century. Nowadays it is involved in the disciplines of psychophysics, cognitive neuroscience and computer science. This paper describes several visual attention models for detecting salient objects in complex scene and focuses on a model based on local context suppression of multiple cues. Although this model is useful to capture visual attention in images containing small objects, it fails in detecting faces as salient objects. For this reason we improved the model by adding more attention cues. We propose a method for detecting salient objects based on texture, where face detection is used as an additional attention cue.