Parametric Model-Based 3D Human Shape and Pose Estimation from Multiple Views
(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)
- author
- Li, Zhongguo LU ; Heyden, Anders LU and Oskarsson, Magnus LU
- organization
- publishing date
- 2019
- 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
- 1611-3349
- 0302-9743
- 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
- 2024-10-02 05:31:12
@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 = {{1611-3349}}, 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}}, }