This paper introduces a new formulation for discrete image

labeling tasks, the Decision Tree Field (DTF), that combines

and generalizes random forests and conditional random

fields (CRF) which have been widely used in computer

vision. In a typical CRF model the unary potentials are derived

from sophisticated random forest or boosting based

classifiers, however, the pairwise potentials are assumed to

(1) have a simple parametric form with a pre-specified and

fixed dependence on the image data, and (2) to be defined on

the basis of a small and fixed neighborhood. In contrast, in

DTF, local interactions between multiple variables are determined

by means of decision trees evaluated on the image

data, allowing the interactions to be adapted to the image

content. This results in powerful graphical models which

are able to represent complex label structure. Our key technical

contribution is to show that the DTF model can be

trained efficiently and jointly using a convex approximate

likelihood function, enabling us to learn over a million free

model parameters. We show experimentally that for applications

which have a rich and complex label structure, our

model achieves excellent results.

PDF file

PDF file

In  ICCV

Details