We consider the task of finding M-best diverse solutions in a graphical model. In a previous work by Batra et al. an algorithmic approach for finding such solutions was proposed, and its usefulness was shown in numerous applications. Contrary to previous work we propose a novel formulation of the problem in form of a single energy minimization problem in a specially constructed graphical model. We show that the method of Batra et al. can be considered as a greedy approximate algorithm for our model, whereas we introduce an efficient specialized optimization technique for it, based on alpha-expansion. We evaluate our method on two application scenarios, interactive and semantic image segmentation, with binary and multiple labels. In both cases we achieve considerably better error rates than state-of-the art diversity methods. Furthermore, we empirically discover that in the binary label case we were able to reach global optimality for all test instances.
%0 Conference Paper
%1 7410568
%A Kirillov, Alexander
%A Savchynskyy, Bogdan
%A Schlesinger, Dmitrij
%A Vetrov, Dmitry
%A Rother, Carsten
%B 2015 IEEE International Conference on Computer Vision (ICCV)
%D 2015
%K Labeling;Graphical methods;Minimization;Computer modeling;Optimization models;Diversity vision;Computational
%P 1814--1822
%R 10.1109/ICCV.2015.211
%T Inferring M-Best Diverse Labelings in a Single One
%X We consider the task of finding M-best diverse solutions in a graphical model. In a previous work by Batra et al. an algorithmic approach for finding such solutions was proposed, and its usefulness was shown in numerous applications. Contrary to previous work we propose a novel formulation of the problem in form of a single energy minimization problem in a specially constructed graphical model. We show that the method of Batra et al. can be considered as a greedy approximate algorithm for our model, whereas we introduce an efficient specialized optimization technique for it, based on alpha-expansion. We evaluate our method on two application scenarios, interactive and semantic image segmentation, with binary and multiple labels. In both cases we achieve considerably better error rates than state-of-the art diversity methods. Furthermore, we empirically discover that in the binary label case we were able to reach global optimality for all test instances.
@inproceedings{7410568,
abstract = {We consider the task of finding M-best diverse solutions in a graphical model. In a previous work by Batra et al. an algorithmic approach for finding such solutions was proposed, and its usefulness was shown in numerous applications. Contrary to previous work we propose a novel formulation of the problem in form of a single energy minimization problem in a specially constructed graphical model. We show that the method of Batra et al. can be considered as a greedy approximate algorithm for our model, whereas we introduce an efficient specialized optimization technique for it, based on alpha-expansion. We evaluate our method on two application scenarios, interactive and semantic image segmentation, with binary and multiple labels. In both cases we achieve considerably better error rates than state-of-the art diversity methods. Furthermore, we empirically discover that in the binary label case we were able to reach global optimality for all test instances.},
added-at = {2024-10-02T10:38:17.000+0200},
author = {Kirillov, Alexander and Savchynskyy, Bogdan and Schlesinger, Dmitrij and Vetrov, Dmitry and Rother, Carsten},
biburl = {https://puma.scadsai.uni-leipzig.de/bibtex/220d317d3c0cd98bb5a69ce4fb318e3e5/scadsfct},
booktitle = {2015 IEEE International Conference on Computer Vision (ICCV)},
doi = {10.1109/ICCV.2015.211},
interhash = {fd90392453ea4e42ef30d73c27698376},
intrahash = {20d317d3c0cd98bb5a69ce4fb318e3e5},
issn = {2380-7504},
keywords = {Labeling;Graphical methods;Minimization;Computer modeling;Optimization models;Diversity vision;Computational},
month = Dec,
pages = {1814--1822},
timestamp = {2024-10-02T10:38:17.000+0200},
title = {Inferring M-Best Diverse Labelings in a Single One},
year = 2015
}
