Computational Visual Attention System
Do you know what a Computational Visual Attention System is?. Visual attention is one among the key mechanisms of perception that permits humans to efficiently choose the visual information of most potential interest. Machines face similar challenges as humans: they need to deal with a large number of input data and need to choose the most promising elements.
Now let us see what attention is and what its need is. Attention is one among the key mechanisms of human perception that permits us to act efficiently in a complicated world. Many data is perceived at every instant, much more than can be processed in detail by the human brain. The power to extract the relevant items of the sensory input at an early processing stage is crucial for efficient acting. Thereby, it depends on the context which a part of the sensory input is relevant.
Computational Visual Attention Systems aim to find regions of interest in pictures. Neurobiologists, psychologists, and computer scientists have investigated visual attention thoroughly during the last decades and profited significantly from one another.
Perhaps the most important outcome of neurophysiological findings on visual attention is that there’s no single brain area guiding the attention. However neural correlates of visual selection seem to be reflected in nearly all brain areas related to visual processing. In addition, new findings indicate that several brain areas share the processing of data from different senses and there’s growing proof; that enormous elements of the cortex are multi sensory.
Some stimuli are intrinsically conspicuous or salient in a given context. When in fact, saliency is independent of the character of the actual task; operates very quickly and is primarily driven in a bottom-up manner, though it is influenced by figure–ground effects. If a stimulus is sufficiently salient, it’ll pop out of a visual scene. This implies that saliency is computed in a pre-attentive manner across the complete visual field; usually in terms of hierarchical centre-surround mechanism. The speed of this saliency-based kind of attention is on the order of 25 to 50 ms per item.
The second type of attention is a more deliberate and powerful one that has variable selection criteria, depending on the task at hand. The expression of this top-down attention is most likely controlled from higher areas, together with the frontal lobes; furthermore that connect back to visual cortex and early visual areas. Such voluntary preparation of attention has a value; because the amount of time that it takes 200 ms or more rivals that is required to move the eyes.0