Perceptual Organization From a Bayesian Point of View

Explore a Bayesian approach to perceptual organization in this lecture by Jacob Feldman from Rutgers University. Delve into the process of how the visual system groups images into distinct units, framing it as an inverse inference problem and mixture estimation. Examine applications in clustering, contour integration, figure/ground estimation, shape representation, part decomposition, object detection, and shape similarity. Discover how this framework unifies diverse grouping rules under the principle of Bayesian posterior maximization, offering a formalization of the Gestalt notion of Prägnanz. Cover topics including Gestalt laws, Bayesian inference, mixtures, history of perceptual grouping, contours, axial models, forward models, part decomposition, and predictive distribution.

Intro
Perceptual Organization
Gestalt Laws
Bayesian Inference
Mixtures
History of perceptual grouping
Contours
Axial Models
Forward Model
Single Axis
Part D Composition
Experiment
Tree Slices
Predictive Distribution