Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Parametric image segmentation of humans with structural shape priors

Popa, Alin Ionut and Sminchisescu, Cristian LU (2017) 13th Asian Conference on Computer Vision, ACCV 2016 In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 10112 LNCS. p.68-83
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... (More)

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.

(Less)
Please use this url to cite or link to this publication:
author
and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
host publication
Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers
series title
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
volume
10112 LNCS
pages
16 pages
publisher
Springer
conference name
13th Asian Conference on Computer Vision, ACCV 2016
conference location
Taipei, Taiwan
conference dates
2016-11-20 - 2016-11-24
external identifiers
  • scopus:85016190750
ISSN
16113349
03029743
ISBN
9783319541839
DOI
10.1007/978-3-319-54184-6_5
language
English
LU publication?
yes
id
391deb84-4f95-4ea1-b71e-a27ed4dcfbc8
date added to LUP
2017-04-06 13:19:37
date last changed
2025-01-07 10:55:37
@inproceedings{391deb84-4f95-4ea1-b71e-a27ed4dcfbc8,
  abstract     = {{<p>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.</p>}},
  author       = {{Popa, Alin Ionut and Sminchisescu, Cristian}},
  booktitle    = {{Computer Vision -  ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers}},
  isbn         = {{9783319541839}},
  issn         = {{16113349}},
  language     = {{eng}},
  pages        = {{68--83}},
  publisher    = {{Springer}},
  series       = {{Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}},
  title        = {{Parametric image segmentation of humans with structural shape priors}},
  url          = {{http://dx.doi.org/10.1007/978-3-319-54184-6_5}},
  doi          = {{10.1007/978-3-319-54184-6_5}},
  volume       = {{10112 LNCS}},
  year         = {{2017}},
}