Automatic Discovery of Meaningful Object Parts with Latent CRFs

Abstract

Object recognition is challenging due to high intra-class variability caused, \ne.g., by articulation, viewpoint changes, and partial occlusion. Successful \nmethods need to strike a balance between being flexible enough to model such \nvariation and discriminative enough to detect objects in cluttered, real world \nscenes. Motivated by these challenges we propose a latent conditional random \nfield (CRF) based on a flexible assembly of parts. By modeling part labels as \nhidden nodes and developing an EM algorithm for learning from class labels \nalone, this new approach enables the automatic discovery of semantically \nmeaningful object part representations. To increase the flexibility and \nexpressiveness of the model, we learn the pairwise structure of the underlying \ngraphical model at the level of object part interactions. Efficient \ngradient-based techniques are used to estimate the structure of the domain of \ninterest and carried forward to the multi-label or object part case. Our \nexperiments illustrate the meaningfulness of the discovered parts and \ndemonstrate state-of-the-art performance of the approach.