DMCA
Author manuscript, published in "PAMI (2010)" 1 OBJCUT: Efficient Segmentation using Top-Down and Bottom-Up Cues (2013)
BibTeX
@MISC{Pawan13authormanuscript,,
author = {M. Pawan and Kumar P. H. S and Torr A. Zisserman},
title = {Author manuscript, published in "PAMI (2010)" 1 OBJCUT: Efficient Segmentation using Top-Down and Bottom-Up Cues},
year = {2013}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
We present a probabilistic method for segmenting instances of a particular object category within an image. Our approach overcomes the deficiencies of previous segmentation techniques based on traditional grid conditional random fields (CRF), namely that (i) they require the user to provide seed pixels for the foreground and the background; and (ii) they provide a poor prior for specific shapes due to the small neighborhood size of grid CRF. Specifically, we automatically obtain the pose of the object in a given image instead of relying on manual interaction. Furthermore, we employ a probabilistic model which includes shape potentials for the object to incorporate top-down information that is global across the image, in addition to the grid clique potentials which provide the bottom-up information used in previous approaches. The shape potentials are provided by the pose of the object obtained using an object category model. We represent articulated object categories using a novel layered pictorial structures model. Nonarticulated object categories are modelled using a set of exemplars. These object category models have the advantage that they can handle large intra-class shape, appearance and spatial variation. We develop an efficient method, OBJCUT, to obtain segmentations using our probabilistic framework. Novel aspects of this method include: (i) efficient algorithms for sampling the object category models of our choice;
Keyphrases
bottom-up cue object category model efficient segmentation author manuscript shape potential object category particular object category grid crf specific shape method include traditional grid conditional random field manual interaction probabilistic framework spatial variation efficient algorithm previous segmentation technique poor prior efficient method novel aspect small neighborhood size probabilistic method bottom-up information top-down information articulated object category grid clique potential probabilistic model pictorial structure model previous approach large intra-class shape seed pixel
