A joint model for 2D and 3D pose estimation from a single image

Edgar Simo Serra, Ariadna Quattoni, Carme Torras, Francesc Moreno-Noguer

Research output: Contribution to journalArticle

58 Citations (Scopus)

Abstract

We introduce a novel approach to automatically recover 3D human pose from a single image. Most previous work follows a pipelined approach: initially, a set of 2D features such as edges, joints or silhouettes are detected in the image, and then these observations are used to infer the 3D pose. Solving these two problems separately may lead to erroneous 3D poses when the feature detector has performed poorly. In this paper, we address this issue by jointly solving both the 2D detection and the 3D inference problems. For this purpose, we propose a Bayesian framework that integrates a generative model based on latent variables and discriminative 2D part detectors based on HOGs, and perform inference using evolutionary algorithms. Real experimentation demonstrates competitive results, and the ability of our methodology to provide accurate 2D and 3D pose estimations even when the 2D detectors are inaccurate.

Original languageEnglish
Article number6619310
Pages (from-to)3634-3641
Number of pages8
JournalUnknown Journal
DOIs
Publication statusPublished - 2013
Externally publishedYes

Fingerprint

Detectors
inference
detectors
experimentation
Evolutionary algorithms
methodology

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition

Cite this

A joint model for 2D and 3D pose estimation from a single image. / Simo Serra, Edgar; Quattoni, Ariadna; Torras, Carme; Moreno-Noguer, Francesc.

In: Unknown Journal, 2013, p. 3634-3641.

Research output: Contribution to journalArticle

Simo Serra, Edgar ; Quattoni, Ariadna ; Torras, Carme ; Moreno-Noguer, Francesc. / A joint model for 2D and 3D pose estimation from a single image. In: Unknown Journal. 2013 ; pp. 3634-3641.
@article{75366e6fdda146ec9ac0e3c80db25bae,
title = "A joint model for 2D and 3D pose estimation from a single image",
abstract = "We introduce a novel approach to automatically recover 3D human pose from a single image. Most previous work follows a pipelined approach: initially, a set of 2D features such as edges, joints or silhouettes are detected in the image, and then these observations are used to infer the 3D pose. Solving these two problems separately may lead to erroneous 3D poses when the feature detector has performed poorly. In this paper, we address this issue by jointly solving both the 2D detection and the 3D inference problems. For this purpose, we propose a Bayesian framework that integrates a generative model based on latent variables and discriminative 2D part detectors based on HOGs, and perform inference using evolutionary algorithms. Real experimentation demonstrates competitive results, and the ability of our methodology to provide accurate 2D and 3D pose estimations even when the 2D detectors are inaccurate.",
author = "{Simo Serra}, Edgar and Ariadna Quattoni and Carme Torras and Francesc Moreno-Noguer",
year = "2013",
doi = "10.1109/CVPR.2013.466",
language = "English",
pages = "3634--3641",
journal = "Nuclear Physics A",
issn = "0375-9474",
publisher = "Elsevier",

}

TY - JOUR

T1 - A joint model for 2D and 3D pose estimation from a single image

AU - Simo Serra, Edgar

AU - Quattoni, Ariadna

AU - Torras, Carme

AU - Moreno-Noguer, Francesc

PY - 2013

Y1 - 2013

N2 - We introduce a novel approach to automatically recover 3D human pose from a single image. Most previous work follows a pipelined approach: initially, a set of 2D features such as edges, joints or silhouettes are detected in the image, and then these observations are used to infer the 3D pose. Solving these two problems separately may lead to erroneous 3D poses when the feature detector has performed poorly. In this paper, we address this issue by jointly solving both the 2D detection and the 3D inference problems. For this purpose, we propose a Bayesian framework that integrates a generative model based on latent variables and discriminative 2D part detectors based on HOGs, and perform inference using evolutionary algorithms. Real experimentation demonstrates competitive results, and the ability of our methodology to provide accurate 2D and 3D pose estimations even when the 2D detectors are inaccurate.

AB - We introduce a novel approach to automatically recover 3D human pose from a single image. Most previous work follows a pipelined approach: initially, a set of 2D features such as edges, joints or silhouettes are detected in the image, and then these observations are used to infer the 3D pose. Solving these two problems separately may lead to erroneous 3D poses when the feature detector has performed poorly. In this paper, we address this issue by jointly solving both the 2D detection and the 3D inference problems. For this purpose, we propose a Bayesian framework that integrates a generative model based on latent variables and discriminative 2D part detectors based on HOGs, and perform inference using evolutionary algorithms. Real experimentation demonstrates competitive results, and the ability of our methodology to provide accurate 2D and 3D pose estimations even when the 2D detectors are inaccurate.

UR - http://www.scopus.com/inward/record.url?scp=84887374930&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84887374930&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2013.466

DO - 10.1109/CVPR.2013.466

M3 - Article

SP - 3634

EP - 3641

JO - Nuclear Physics A

JF - Nuclear Physics A

SN - 0375-9474

M1 - 6619310

ER -