Joint training of a convolutional network and a graphical model for human pose estimation

Jonathan Tompson; Arjun Jain; Yann LeCun; Christoph Bregler

Joint training of a convolutional network and a graphical model for human pose estimation

Jonathan Tompson, Arjun Jain, Yann LeCun, Christoph Bregler

Computer Science

Research output: Contribution to journal › Conference article › peer-review

Abstract

This paper proposes a new hybrid architecture that consists of a deep Convolutional Network and a Markov Random Field. We show how this architecture is successfully applied to the challenging problem of articulated human pose estimation in monocular images. The architecture can exploit structural domain constraints such as geometric relationships between body joint locations. We show that joint training of these two model paradigms improves performance and allows us to significantly outperform existing state-of-the-art techniques.

Original language	English (US)
Pages (from-to)	1799-1807
Number of pages	9
Journal	Advances in Neural Information Processing Systems
Volume	2
Issue number	January
State	Published - 2014
Event	28th Annual Conference on Neural Information Processing Systems 2014, NIPS 2014 - Montreal, Canada Duration: Dec 8 2014 → Dec 13 2014

ASJC Scopus subject areas

Computer Networks and Communications
Information Systems
Signal Processing

Cite this

@article{467a64cc72e249ffac0bd706929ab3a3,

title = "Joint training of a convolutional network and a graphical model for human pose estimation",

abstract = "This paper proposes a new hybrid architecture that consists of a deep Convolutional Network and a Markov Random Field. We show how this architecture is successfully applied to the challenging problem of articulated human pose estimation in monocular images. The architecture can exploit structural domain constraints such as geometric relationships between body joint locations. We show that joint training of these two model paradigms improves performance and allows us to significantly outperform existing state-of-the-art techniques.",

author = "Jonathan Tompson and Arjun Jain and Yann LeCun and Christoph Bregler",

year = "2014",

language = "English (US)",

volume = "2",

pages = "1799--1807",

journal = "Advances in Neural Information Processing Systems",

issn = "1049-5258",

publisher = "Neural information processing systems foundation",

number = "January",

note = "28th Annual Conference on Neural Information Processing Systems 2014, NIPS 2014 ; Conference date: 08-12-2014 Through 13-12-2014",

}

TY - JOUR

T1 - Joint training of a convolutional network and a graphical model for human pose estimation

AU - Tompson, Jonathan

AU - Jain, Arjun

AU - LeCun, Yann

AU - Bregler, Christoph

PY - 2014

Y1 - 2014

N2 - This paper proposes a new hybrid architecture that consists of a deep Convolutional Network and a Markov Random Field. We show how this architecture is successfully applied to the challenging problem of articulated human pose estimation in monocular images. The architecture can exploit structural domain constraints such as geometric relationships between body joint locations. We show that joint training of these two model paradigms improves performance and allows us to significantly outperform existing state-of-the-art techniques.

AB - This paper proposes a new hybrid architecture that consists of a deep Convolutional Network and a Markov Random Field. We show how this architecture is successfully applied to the challenging problem of articulated human pose estimation in monocular images. The architecture can exploit structural domain constraints such as geometric relationships between body joint locations. We show that joint training of these two model paradigms improves performance and allows us to significantly outperform existing state-of-the-art techniques.

UR - http://www.scopus.com/inward/record.url?scp=84930634156&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84930634156&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:84930634156

SN - 1049-5258

VL - 2

SP - 1799

EP - 1807

JO - Advances in Neural Information Processing Systems

JF - Advances in Neural Information Processing Systems

IS - January

T2 - 28th Annual Conference on Neural Information Processing Systems 2014, NIPS 2014

Y2 - 8 December 2014 through 13 December 2014

ER -

Joint training of a convolutional network and a graphical model for human pose estimation

Abstract

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this