Computing with connections in visual recognition of origami objects

This poper summarizes our initial foray in tackling Artificial Intelligence problems using o connectionist approach. The particular task chosen was the visual recognition of objects in the Origami world OS defined by Kanade (1978). The two major questions answered were how to construct o connectionist network to represent and recognize projected line drawings of Origami objects and what advantages such on approach would have. The structure of the resulting connectionist network con be described OS o hierarchy of parameter or feature spaces with each node in each of the feature spaces representing o hypothesis about the possible existence of a specific geometric feature of on Origami object. The dynamic behavior of the network is o form of iterative refinement or relaxation whose major characteristic is to prefer more globally interesting interpretations of the input over locally pleasing ones. Examples from the implementation illustrate the system’s ability to deal with forms of noise, occlusion and missing information. Other benefits ore on inherently parallel approach to vision, limitation of explicit ordering of the search involved in matching model to instance and the elimination of backtracking due to the shoring of partial results OS the search progresses. Extensions and problems ore also discussed.