Decision tree fields

Authors:
Sebastian Nowozin;Carsten Rother;Shai Bagon;Toby Sharp; Bangpeng Yao;Pushmeet Kohli
Affiliations:
Microsoft Research, Cambridge, UK;Microsoft Research, Cambridge, UK;Weizmann Institute, Israel;Microsoft Research, Cambridge, UK;Stanford University, CA, USA;Microsoft Research, Cambridge, UK
Venue:
ICCV '11 Proceedings of the 2011 International Conference on Computer Vision
Year:
2011

Citing 0
Cited 6

Decision Forests: A Unified Framework for Classification, Regression, Density Estimation, Manifold Learning and Semi-Supervised Learning

Foundations and Trends® in Computer Graphics and Vision
Learning discriminative spatial relations for detector dictionaries: an application to pedestrian detection

ECCV'12 Proceedings of the 12th European conference on Computer Vision - Volume Part II
Loss-specific training of non-parametric image restoration models: a new state of the art

ECCV'12 Proceedings of the 12th European conference on Computer Vision - Volume Part VII
The lazy flipper: efficient depth-limited exhaustive search in discrete graphical models

ECCV'12 Proceedings of the 12th European conference on Computer Vision - Volume Part VII
Joint classification-regression forests for spatially structured multi-object segmentation

ECCV'12 Proceedings of the 12th European conference on Computer Vision - Volume Part IV
Superparsing

International Journal of Computer Vision

Quantified Score

Hi-index	0.00

Visualization

Abstract

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.