Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Parametric Model-Based 3D Human Shape and Pose Estimation from Multiple Views

Li, Zhongguo LU ; Heyden, Anders LU orcid and Oskarsson, Magnus LU orcid (2019) 21st Scandinavian Conference on Image Analysis, SCIA 2019 In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 11482 LNCS. p.336-347
Abstract

Human body pose and shape estimation is an important and challenging task in computer vision. This paper presents a novel method for estimating 3D human body pose and shape from several RGB images, using detected joint positions in the images and based on a parametric human body model. Firstly, the 2D joint points of the RGB images are estimated using a deep neural network, which provides a strong prior on the pose. Then, an energy function is constructed based on the 2D joint points in the RGB images and a parametric human body model. By minimizing the energy function, the pose, shape and camera parameters are obtained. The main contribution of the method over previous work, is that the optimization is based on several images... (More)

Human body pose and shape estimation is an important and challenging task in computer vision. This paper presents a novel method for estimating 3D human body pose and shape from several RGB images, using detected joint positions in the images and based on a parametric human body model. Firstly, the 2D joint points of the RGB images are estimated using a deep neural network, which provides a strong prior on the pose. Then, an energy function is constructed based on the 2D joint points in the RGB images and a parametric human body model. By minimizing the energy function, the pose, shape and camera parameters are obtained. The main contribution of the method over previous work, is that the optimization is based on several images simultaneously using only estimated joint positions in the images. We have performed experiments on both synthetic and real image data-sets, that demonstrate that our method can reconstruct 3D human bodies with better accuracy than previous single view methods.

(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
keywords
Camera, Human body, Parametric model, Pose and shape estimation, RGB images
host publication
Image Analysis - 21st Scandinavian Conference, SCIA 2019, Proceedings
series title
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
editor
Felsberg, Michael ; Forssén, Per-Erik ; Unger, Jonas and Sintorn, Ida-Maria
volume
11482 LNCS
pages
12 pages
publisher
Springer
conference name
21st Scandinavian Conference on Image Analysis, SCIA 2019
conference location
Norrköping, Sweden
conference dates
2019-06-11 - 2019-06-13
external identifiers
  • scopus:85066912400
ISSN
0302-9743
1611-3349
ISBN
9783030202040
DOI
10.1007/978-3-030-20205-7_28
language
English
LU publication?
yes
id
99328ca1-5892-473a-89d6-af54ee8a2159
date added to LUP
2019-06-19 14:19:43
date last changed
2024-03-03 15:35:23
@inproceedings{99328ca1-5892-473a-89d6-af54ee8a2159,
  abstract     = {{<p>Human body pose and shape estimation is an important and challenging task in computer vision. This paper presents a novel method for estimating 3D human body pose and shape from several RGB images, using detected joint positions in the images and based on a parametric human body model. Firstly, the 2D joint points of the RGB images are estimated using a deep neural network, which provides a strong prior on the pose. Then, an energy function is constructed based on the 2D joint points in the RGB images and a parametric human body model. By minimizing the energy function, the pose, shape and camera parameters are obtained. The main contribution of the method over previous work, is that the optimization is based on several images simultaneously using only estimated joint positions in the images. We have performed experiments on both synthetic and real image data-sets, that demonstrate that our method can reconstruct 3D human bodies with better accuracy than previous single view methods.</p>}},
  author       = {{Li, Zhongguo and Heyden, Anders and Oskarsson, Magnus}},
  booktitle    = {{Image Analysis - 21st Scandinavian Conference, SCIA 2019, Proceedings}},
  editor       = {{Felsberg, Michael and Forssén, Per-Erik and Unger, Jonas and Sintorn, Ida-Maria}},
  isbn         = {{9783030202040}},
  issn         = {{0302-9743}},
  keywords     = {{Camera; Human body; Parametric model; Pose and shape estimation; RGB images}},
  language     = {{eng}},
  pages        = {{336--347}},
  publisher    = {{Springer}},
  series       = {{Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}},
  title        = {{Parametric Model-Based 3D Human Shape and Pose Estimation from Multiple Views}},
  url          = {{http://dx.doi.org/10.1007/978-3-030-20205-7_28}},
  doi          = {{10.1007/978-3-030-20205-7_28}},
  volume       = {{11482 LNCS}},
  year         = {{2019}},
}