Recognition and learning with polymorphic structural components

We address the problem of describing recognizing and learning generic free form objects in real world scenes The appearance based system rep resents objects in a graph form and uses weak structure and evidence accu mulation to implicitly encode object models in a learned decision tree after the method developed by Bischof and Caelli Pattern Recognition pp The decision tree is used to classify sequences of im age components or part paths extracted from the object to be recognized The part paths are in turn used to accumulate evidence for the classi ca tion of the entire object We introduce an improved method for generating part paths based upon the Part Compatibility Graph PCG a replacement for Bischof and Caelli s implicit use of the Part Adjacency Graph PAG A new formalism for extending the representation and recognition scheme to utilize multiple polymorphic types of primitive parts is presented and the approach is demonstrated on a selection of imagery