To date, there is no unequivocal evidence for separate visual computations of object features for performance of goal-directed actions versus perceptual tasks in the neurologically intact human brain.
During each visual fixation, complex visual computations and the preparation for the next eye movement are accomplished in as little as 120 ms (Kirchner and Thorpe 2006).
Major arguments for the definition of a visual front end are that the first stages of visual processing should be as uncommitted as possible and allow initial processing steps to be shared between different later-stage visual modules, thus implying a uniform structure on the first stages of visual computations (Koenderink et al. 1992; Lindeberg 1994b, Sect. 1.1).
These findings have prompted the theory that visual computations depends both on bottom-up sensory input and on contextual signals that bias perception toward a particular interpretation of the visual world (Mumford 1992; Rao and Ballard 1999; Bar 2004; Friston 2005; Gilbert and Sigman 2007).
This control condition was necessary in order to ensure that reduced performance from masking did not simply reflect the fact that the targets were effectively invisible, but revealed limits on the visual computations within the sensory image aimed to arrive at a structured, object-like representation.
Salient region detection is different from the visual saliency computation in [10, 36] or other based on the biologically plausible computational models of attention.
To reach real-time, the operations required for visual attention computation were optimally distributed on the analog and digital parts.
With generalisation to other tasks in mind, these detectors are presented as instances of a generic neural-swarm layered architecture for visual saliency computation.
Herein, we propose an extension of our baseline multimodal saliency-frontend (TMM13) [12], which is a unified energy-based framework for audio-visual saliency computation.
