Generic Cuts: An Efficient Algorithm for Optimal Inference in Higher Order MRF-MAP

Published on Feb 4, 20254243 Views

Chetan Arora

We propose a new algorithm called Generic Cuts for computing optimal solutions to 2 label MRF-MAP problems with higher order clique potentials satisfying submodularity. The algorithm runs in time O(2

ECCV 2012 - Firenze

Related categories

Presentation

GENERIC CUTS: AN EFFICIENT ALGORITHM FOR OPTIMAL INFERENCE IN HIGHER ORDER MRF-MAP00:00

MRF-MAP INFERENCE (1)03:50:53

MRF-MAP INFERENCE (2)12:08:16

HIGHER ORDER MRF – WHY?17:11:12

POSSIBLE INFERENCE METHODS31:23:37

MAIN CONTRIBUTION65:54:25

GADGET82:33:41

EDGE CAPACITY – DUAL FEASIBILITY CONSTRAINT (1)101:40:25

EDGE CAPACITY – DUAL FEASIBILITY CONSTRAINT (2)116:50:42

HOW DOES SUBMODULARITY EFFECTS US - MAXFLOW?137:39:42

HOW DOES SUBMODULARITY EFFECTS US - MINCUT?147:13:30

COMBINATORIAL PROPERTIES OF THE FRAMEWORK170:57:53

COMPARISON195:22:39

COMPARISON -ENERGY , CLIQUE SIZE=4213:43:48

COMPARISON -TIME (MS) , CLIQUE SIZE=4221:58:28

COMPARISON – CLIQUE SIZE V/S TIME (MS)224:23:02

GENERALIZATIONS232:46:57

SUMMARY236:40:01