High-Level Perception and Low-Level Vision: Bridging the Semantic Gap

The "semantic gap" describes the current inability of computer vision systems to connect low-level visual descriptions with the high-level conceptual schemas and analogies. Similarly, neuroscientists and psychophysicists have uncovered many of the physiological and perceptual mechanisms underlying various stages of visual processing. Yet the interactions and transformations between levels remains obscure. The purpose of this workshop was to address how the semantic gap might be bridged, both in natural and computer vision, and to see if a set of common principles of visual understanding can be discovered from the collective, multidisciplinary knowledge of the participants.