MPLP++: Fast, Parallel Dual Block-Coordinate Ascent for Dense Graphical Models

04/16/2020
by   Siddharth Tourani, et al.
0

Dense, discrete Graphical Models with pairwise potentials are a powerful class of models which are employed in state-of-the-art computer vision and bio-imaging applications. This work introduces a new MAP-solver, based on the popular Dual Block-Coordinate Ascent principle. Surprisingly, by making a small change to the low-performing solver, the Max Product Linear Programming (MPLP) algorithm, we derive the new solver MPLP++ that significantly outperforms all existing solvers by a large margin, including the state-of-the-art solver Tree-Reweighted Sequential (TRWS) message-passing algorithm. Additionally, our solver is highly parallel, in contrast to TRWS, which gives a further boost in performance with the proposed GPU and multi-thread CPU implementations. We verify the superiority of our algorithm on dense problems from publicly available benchmarks, as well, as a new benchmark for 6D Object Pose estimation. We also provide an ablation study with respect to graph density.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
04/16/2020

Taxonomy of Dual Block-Coordinate Ascent Methods for Discrete Energy Minimization

We consider the maximum-a-posteriori inference problem in discrete graph...
research
08/31/2015

Maximum Persistency via Iterative Relaxed Inference with Graphical Models

We consider the NP-hard problem of MAP-inference for undirected discrete...
research
12/16/2016

A Study of Lagrangean Decompositions and Dual Ascent Solvers for Graph Matching

We study the quadratic assignment problem, in computer vision also known...
research
12/04/2020

Efficient semidefinite-programming-based inference for binary and multi-class MRFs

Probabilistic inference in pairwise Markov Random Fields (MRFs), i.e. co...
research
12/06/2013

Dual coordinate solvers for large-scale structural SVMs

This manuscript describes a method for training linear SVMs (including b...
research
11/29/2016

Efficient Linear Programming for Dense CRFs

The fully connected conditional random field (CRF) with Gaussian pairwis...
research
08/24/2021

Making Higher Order MOT Scalable: An Efficient Approximate Solver for Lifted Disjoint Paths

We present an efficient approximate message passing solver for the lifte...

Please sign up or login with your details

Forgot password? Click here to reset