Natural scene categorization from the conjunction of ecological global properties.

Proceedings of the Twenty-Eighth Annual Conference of the Cognitive Science Society, 291-296.

Greene, M.R., & Oliva, A. (2006).

Human scene understanding is remarkable: with only a brief glance at an image, an abundance of information is available -spatial layout, scene function, semantic label, etc. Here we propose a scene-centered model of rapid human scene understanding that uses a vocabulary of global, ecological scene properties that combine to categorize natural landscape images.¾ Behaviorally, we show human observers are sensitive to the underlying distributions of these global properties for use in basic-level categorization.¾ An ideal observer trained only on the distributions of these properties predicts human scene categorization performance (r=0.90) and human errors.

articleposter