The Structure of Visual Content

Gabe Greenberg, UCLA


This will be a workshop-style discussion of some new ideas in the semantic analysis of visual representation. I’ll focus on the use of "feature maps" to characterize a wide range of visual content, with an eye towards specifying precise accuracy conditions. The primary application will be static pictorial images (such as drawings or photographs), but we’ll also consider extensions to dynamic displays such as comics, film, and iconicity in gesture and sign language. As time permits, we’ll discuss a range of semi-technical issues that arise for feature map semantics, including the representation of angle, depth, category, quantification, transparency, relations, and temporal duration.