Abstract

We present a novel method for unsupervised classification, including the discovery of a new category and precise ob ject and part localization. Given a set of unlabelled images, some of which contain an ob ject of an unknown category, with unknown location and unknown size relative to the background, the method automatically identifies the im- ages that contain the ob jects, localizes them and their parts, and reliably learns their appearance and geometry for subsequent classification. Cur- rent unsupervised methods construct classifiers based on a fixed set of initial features. Instead, we propose a new approach which iteratively ex- tracts new features and re-learns the induced classifier, improving class vs. non-class separation at each iteration. We develop two main tools that allow this iterative combined search. The first is a novel star-like model capable of learning a geometric class representation in the unsu- pervised setting. The second is learning of ”part specific features” that are optimized for parts detection, and which optimally combine differ- ent part appearances discovered in the training examples. These novel aspects lead to precise part localization and to improvement in overall classification performance compared with previous methods. We applied our method to multiple ob ject classes from Caltech-101, UIUC and a sub-classification problem from PASCAL. The obtained results are com- parable to state-of-the-art supervised classification techniques and su- perior to state-of-the-art unsupervised approaches previously applied to the same image sets.