Unsupervised learning of categorical segments in image collections

Marco Andreetto, Lihi Zelnik-Manor, Pietro Perona

Abstract: Which one comes first: segmentation or recognition? We propose a probabilistic framework for carrying out the two simultaneously. The framework combines an LDA dasiabag of visual wordspsila model for recognition, and a hybrid parametric-nonparametric model for segmentation. If applied to a collection of images, our framework can simultaneously discover the segments of each image, and the correspondence between such segments. Such segments may be thought of as the dasiapartspsila of corresponding objects that appear in the image collection. Thus, the model may be used for learning new categories, detecting/classifying objects, and segmenting images.