| | |
Summary: Unsupervised Learning of Categorical Segments in Image Collections
Marco Andreetto
Lihi Zelnik-Manor
Pietro Perona
Dept. of Electrical Engineering
Dept. of Electrical Engineering
California Institute of Technology Technion - Israel Institute of Technology
Pasadena, CA 91125, United States Haifa, 32000, Israel
Abstract
Which one comes first: segmentation or recognition? We
propose a probabilistic framework for carrying out the
two simultaneously. The framework combines an LDA
`bag of visual words' model for recognition, and a hybrid
parametric-nonparametric model for segmentation. If ap-
plied to a collection of images, our framework can simulta-
neously discover the segments of each image, and the corre-
spondence between such segments. Such segments may be
thought of as the `parts' of corresponding objects that ap-
pear in the image collection. Thus, the model may be used
|
