Abstract
This paper introduces the Located Hidden Random Field (LHRF), a conditional model for simultaneous part-based detection and segmentation of ob jects of a given class. Given a training set of images with segmentation masks for the ob ject of interest, the LHRF automati- cally learns a set of parts that are both discriminative in terms of appear- ance and informative about the location of the ob ject. By introducing the global position of the ob ject as a latent variable, the LHRF models the long-range spatial configuration of these parts, as well as their local interactions. Experiments on benchmark datasets show that the use of discriminative parts leads to state-of-the-art detection and segmentation performance, with the additional benefit of obtaining a labeling of the ob ject’s component parts.