Improving Convergence in Hierarchical Matching Networks for Object Recognition

Abstract

We are interested in the use of analog neural networks for recog(cid:173) nizing visual objects. Objects are described by the set of parts they are composed of and their structural relationship. Struc(cid:173) tural models are stored in a database and the recognition prob(cid:173) lem reduces to matching data to models in a structurally consis(cid:173) tent way. The object recognition problem is in general very diffi(cid:173) cult in that it involves coupled problems of grouping, segmentation and matching. We limit the problem here to the simultaneous la(cid:173) belling of the parts of a single object and the determination of analog parameters. This coupled problem reduces to a weighted match problem in which an optimizing neural network must min(cid:173) imize E(M, p) = LO'i MO'i WO'i(p), where the {MO'd are binary match variables for data parts i to model parts a and Wai(P) are weights dependent on parameters p . In this work we show that by first solving for estimates p without solving for M ai , we may obtain good initial parameter estimates that yield better solutions for M and p.