Parametric image segmentation of humans with structural shape priors

Alin Ionut Popa, Cristian Sminchisescu

Research output: Chapter in Book/Report/Conference proceedingPaper in conference proceedingpeer-review

Abstract

The figure-ground segmentation of humans in images captured in natural environments is an outstanding open problem due to the presence of complex backgrounds, articulation, varying body proportions, partial views and viewpoint changes. In this work we propose classspecific segmentation models that leverage parametric max-flow image segmentation and a large dataset of human shapes. Our contributions are as follows: (1) formulation of a submodular energy model that combines classspecific structural constraints and datadriven shape priors, within a parametric max-flow optimization methodology that systematically computes all breakpoints of the model in polynomial time; (2) design of a datadriven classspecific fusion methodology, based on matching against a large training set of exemplar human shapes (100,000 in our experiments), that allows the shape prior to be constructed on-the-fly, for arbitrary viewpoints and partial views.

Original languageEnglish
Title of host publicationComputer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers
PublisherSpringer
Pages68-83
Number of pages16
Volume10112 LNCS
ISBN (Print)9783319541839
DOIs
Publication statusPublished - 2017
Event13th Asian Conference on Computer Vision, ACCV 2016 - Taipei, Taiwan
Duration: 2016 Nov 202016 Nov 24

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10112 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Conference

Conference13th Asian Conference on Computer Vision, ACCV 2016
Country/TerritoryTaiwan
City Taipei
Period2016/11/202016/11/24

Subject classification (UKÄ)

  • Computer Vision and Robotics (Autonomous Systems)

Fingerprint

Dive into the research topics of 'Parametric image segmentation of humans with structural shape priors'. Together they form a unique fingerprint.

Cite this