Object Recognition by Combining Appearance and Geometry

We present a new class of statistical models for part-based object recognition. These models are explicitly parametrized according to the degree of spatial structure that they can represent. This provides a way of relating different spatial priors that have been used in the past such as joint Gaussian models and tree-structured models. By providing explicit control over the degree of spatial structure, our models make it possible to study questions such as the extent to which additional spatial constraints among parts are helpful in detection and localization, and the tradeoff between representational power and computational cost. We consider these questions for object classes that have substantial geometric structure, such as airplanes, faces and motorbikes, using datasets employed by other researchers to facilitate evaluation. We find that for these classes of objects, a relatively small amount of spatial structure in the model can provide statistically indistinguishable recognition performance from more powerful models, and at a substantially lower computational cost.