Kernel conditional ordinal random fields for temporal segmentation of facial action units

Ognjen Rudovic; Vladimir Pavlovic; Maja Pantic

doi:10.1007/978-3-642-33868-7_26

Kernel conditional ordinal random fields for temporal segmentation of facial action units

Ognjen Rudovic, Vladimir Pavlovic, Maja Pantic

Institute for Quantitative Biomedicine (IQB)

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

18 Scopus citations

Abstract

We consider the problem of automated recognition of temporal segments (neutral, onset, apex and offset) of Facial Action Units. To this end, we propose the Laplacian-regularized Kernel Conditional Ordinal Random Field model. In contrast to standard modeling approaches to recognition of AUs' temporal segments, which treat each segment as an independent class, the proposed model takes into account ordinal relations between the segments. The experimental results evidence the effectiveness of such an approach.

Original language	English (US)
Title of host publication	Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings
Publisher	Springer Verlag
Pages	260-269
Number of pages	10
Edition	PART 2
ISBN (Print)	9783642338670
DOIs	https://doi.org/10.1007/978-3-642-33868-7_26
State	Published - 2012
Event	Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings - Florence, Italy Duration: Oct 7 2012 → Oct 13 2012

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 2
Volume	7584 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings
Country/Territory	Italy
City	Florence
Period	10/7/12 → 10/13/12

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Keywords

Action units
conditional random field
histogram intersection kernel
kernel locality preserving projections
ordinal regression

Access to Document

10.1007/978-3-642-33868-7_26

Cite this

Rudovic, O., Pavlovic, V., & Pantic, M. (2012). Kernel conditional ordinal random fields for temporal segmentation of facial action units. In Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings (PART 2 ed., pp. 260-269). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7584 LNCS, No. PART 2). Springer Verlag. https://doi.org/10.1007/978-3-642-33868-7_26

Rudovic, Ognjen ; Pavlovic, Vladimir ; Pantic, Maja. / Kernel conditional ordinal random fields for temporal segmentation of facial action units. Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings. PART 2. ed. Springer Verlag, 2012. pp. 260-269 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).

@inproceedings{0864b44b8de44a429576d5981b762874,

title = "Kernel conditional ordinal random fields for temporal segmentation of facial action units",

abstract = "We consider the problem of automated recognition of temporal segments (neutral, onset, apex and offset) of Facial Action Units. To this end, we propose the Laplacian-regularized Kernel Conditional Ordinal Random Field model. In contrast to standard modeling approaches to recognition of AUs' temporal segments, which treat each segment as an independent class, the proposed model takes into account ordinal relations between the segments. The experimental results evidence the effectiveness of such an approach.",

keywords = "Action units, conditional random field, histogram intersection kernel, kernel locality preserving projections, ordinal regression",

author = "Ognjen Rudovic and Vladimir Pavlovic and Maja Pantic",

year = "2012",

doi = "10.1007/978-3-642-33868-7_26",

language = "English (US)",

isbn = "9783642338670",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

number = "PART 2",

pages = "260--269",

booktitle = "Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings",

address = "Germany",

edition = "PART 2",

note = "Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings ; Conference date: 07-10-2012 Through 13-10-2012",

}

Rudovic, O, Pavlovic, V & Pantic, M 2012, Kernel conditional ordinal random fields for temporal segmentation of facial action units. in Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings. PART 2 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 7584 LNCS, Springer Verlag, pp. 260-269, Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings, Florence, Italy, 10/7/12. https://doi.org/10.1007/978-3-642-33868-7_26

Kernel conditional ordinal random fields for temporal segmentation of facial action units. / Rudovic, Ognjen; Pavlovic, Vladimir; Pantic, Maja.
Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings. PART 2. ed. Springer Verlag, 2012. p. 260-269 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7584 LNCS, No. PART 2).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Kernel conditional ordinal random fields for temporal segmentation of facial action units

AU - Rudovic, Ognjen

AU - Pavlovic, Vladimir

AU - Pantic, Maja

PY - 2012

Y1 - 2012

N2 - We consider the problem of automated recognition of temporal segments (neutral, onset, apex and offset) of Facial Action Units. To this end, we propose the Laplacian-regularized Kernel Conditional Ordinal Random Field model. In contrast to standard modeling approaches to recognition of AUs' temporal segments, which treat each segment as an independent class, the proposed model takes into account ordinal relations between the segments. The experimental results evidence the effectiveness of such an approach.

AB - We consider the problem of automated recognition of temporal segments (neutral, onset, apex and offset) of Facial Action Units. To this end, we propose the Laplacian-regularized Kernel Conditional Ordinal Random Field model. In contrast to standard modeling approaches to recognition of AUs' temporal segments, which treat each segment as an independent class, the proposed model takes into account ordinal relations between the segments. The experimental results evidence the effectiveness of such an approach.

KW - Action units

KW - conditional random field

KW - histogram intersection kernel

KW - kernel locality preserving projections

KW - ordinal regression

UR - http://www.scopus.com/inward/record.url?scp=84867715757&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867715757&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-33868-7_26

DO - 10.1007/978-3-642-33868-7_26

M3 - Conference contribution

AN - SCOPUS:84867715757

SN - 9783642338670

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 260

EP - 269

BT - Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings

PB - Springer Verlag

T2 - Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings

Y2 - 7 October 2012 through 13 October 2012

ER -

Rudovic O, Pavlovic V, Pantic M. Kernel conditional ordinal random fields for temporal segmentation of facial action units. In Computer Vision, ECCV 2012 - Workshops and Demonstrations, Proceedings. PART 2 ed. Springer Verlag. 2012. p. 260-269. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). doi: 10.1007/978-3-642-33868-7_26

Kernel conditional ordinal random fields for temporal segmentation of facial action units

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Keywords

Access to Document

Other files and links

Fingerprint

Cite this