Adversarial Feature Distillation for Facial Expression Recognition

Mengchao Bai; Xi Jia; Weicheng Xie; Linlin Shen

doi:10.1007/978-3-030-29894-4_7

Adversarial Feature Distillation for Facial Expression Recognition

Mengchao Bai, Xi Jia, Weicheng Xie, Linlin Shen

Research output: Chapter in Book/Conference proceeding › Conference contribution › peer-review

Abstract

Human face image contains abundant information including expression, age and gender, etc. Therefore, extracting discriminative feature for certain attribute while expelling others is critical for single facial attribute analysis. In this paper, we propose an adversarial facial expression recognition system, named expression distilling and dispelling learning (ED (formula presented)L), to extract discriminative expression feature from a given face image. The proposed ED (formula presented)L framework composed of two branches, i.e. expression distilling branch ED (formula presented)L-t and expression dispelling branch ED (formula presented)L-p. The ED (formula presented)L-t branch aims to extract the expression-related feature, while the ED (formula presented)L-p branch extracts the non-related feature. The disentangled features jointly serve as a complete representation of the face. Extensive experiments on several benchmark databases, i.e. the CK+, MMI, BU-3DFE and Oulu-CASIA, demonstrate the effectiveness of the proposed ED (formula presented)L framework.

Original language	English
Title of host publication	PRICAI 2019
Subtitle of host publication	Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings
Editors	Abhaya C. Nayak, Alok Sharma
Publisher	Springer Verlag
Pages	80-92
Number of pages	13
ISBN (Print)	9783030298937
DOIs	https://doi.org/10.1007/978-3-030-29894-4_7
Publication status	Published - 2019
Externally published	Yes
Event	16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019 - Yanuka Island, Fiji Duration: 26 Aug 2019 → 30 Aug 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11672 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019
Country/Territory	Fiji
City	Yanuka Island
Period	26/08/19 → 30/08/19

Keywords

Adversarial learning
Facial expression recognition
Feature dispelling
Feature distilling

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-030-29894-4_7

Cite this

Bai, M., Jia, X., Xie, W., & Shen, L. (2019). Adversarial Feature Distillation for Facial Expression Recognition. In A. C. Nayak, & A. Sharma (Eds.), PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings (pp. 80-92). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11672 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-030-29894-4_7

Bai, Mengchao ; Jia, Xi ; Xie, Weicheng et al. / Adversarial Feature Distillation for Facial Expression Recognition. PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. editor / Abhaya C. Nayak ; Alok Sharma. Springer Verlag, 2019. pp. 80-92 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{259e79a932e34a0984154acf4a2c0d15,

title = "Adversarial Feature Distillation for Facial Expression Recognition",

abstract = "Human face image contains abundant information including expression, age and gender, etc. Therefore, extracting discriminative feature for certain attribute while expelling others is critical for single facial attribute analysis. In this paper, we propose an adversarial facial expression recognition system, named expression distilling and dispelling learning (ED (formula presented)L), to extract discriminative expression feature from a given face image. The proposed ED (formula presented)L framework composed of two branches, i.e. expression distilling branch ED (formula presented)L-t and expression dispelling branch ED (formula presented)L-p. The ED (formula presented)L-t branch aims to extract the expression-related feature, while the ED (formula presented)L-p branch extracts the non-related feature. The disentangled features jointly serve as a complete representation of the face. Extensive experiments on several benchmark databases, i.e. the CK+, MMI, BU-3DFE and Oulu-CASIA, demonstrate the effectiveness of the proposed ED (formula presented)L framework.",

keywords = "Adversarial learning, Facial expression recognition, Feature dispelling, Feature distilling",

author = "Mengchao Bai and Xi Jia and Weicheng Xie and Linlin Shen",

note = "Publisher Copyright: {\textcopyright} 2019, Springer Nature Switzerland AG.; 16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019 ; Conference date: 26-08-2019 Through 30-08-2019",

year = "2019",

doi = "10.1007/978-3-030-29894-4_7",

language = "English",

isbn = "9783030298937",

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

publisher = "Springer Verlag",

pages = "80--92",

editor = "Nayak, {Abhaya C.} and Alok Sharma",

booktitle = "PRICAI 2019",

address = "Germany",

}

Bai, M, Jia, X, Xie, W & Shen, L 2019, Adversarial Feature Distillation for Facial Expression Recognition. in AC Nayak & A Sharma (eds), PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11672 LNAI, Springer Verlag, pp. 80-92, 16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019, Yanuka Island, Fiji, 26/08/19. https://doi.org/10.1007/978-3-030-29894-4_7

Adversarial Feature Distillation for Facial Expression Recognition. / Bai, Mengchao; Jia, Xi; Xie, Weicheng et al.
PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. ed. / Abhaya C. Nayak; Alok Sharma. Springer Verlag, 2019. p. 80-92 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11672 LNAI).

Research output: Chapter in Book/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Adversarial Feature Distillation for Facial Expression Recognition

AU - Bai, Mengchao

AU - Jia, Xi

AU - Xie, Weicheng

AU - Shen, Linlin

PY - 2019

Y1 - 2019

N2 - Human face image contains abundant information including expression, age and gender, etc. Therefore, extracting discriminative feature for certain attribute while expelling others is critical for single facial attribute analysis. In this paper, we propose an adversarial facial expression recognition system, named expression distilling and dispelling learning (ED (formula presented)L), to extract discriminative expression feature from a given face image. The proposed ED (formula presented)L framework composed of two branches, i.e. expression distilling branch ED (formula presented)L-t and expression dispelling branch ED (formula presented)L-p. The ED (formula presented)L-t branch aims to extract the expression-related feature, while the ED (formula presented)L-p branch extracts the non-related feature. The disentangled features jointly serve as a complete representation of the face. Extensive experiments on several benchmark databases, i.e. the CK+, MMI, BU-3DFE and Oulu-CASIA, demonstrate the effectiveness of the proposed ED (formula presented)L framework.

AB - Human face image contains abundant information including expression, age and gender, etc. Therefore, extracting discriminative feature for certain attribute while expelling others is critical for single facial attribute analysis. In this paper, we propose an adversarial facial expression recognition system, named expression distilling and dispelling learning (ED (formula presented)L), to extract discriminative expression feature from a given face image. The proposed ED (formula presented)L framework composed of two branches, i.e. expression distilling branch ED (formula presented)L-t and expression dispelling branch ED (formula presented)L-p. The ED (formula presented)L-t branch aims to extract the expression-related feature, while the ED (formula presented)L-p branch extracts the non-related feature. The disentangled features jointly serve as a complete representation of the face. Extensive experiments on several benchmark databases, i.e. the CK+, MMI, BU-3DFE and Oulu-CASIA, demonstrate the effectiveness of the proposed ED (formula presented)L framework.

KW - Adversarial learning

KW - Facial expression recognition

KW - Feature dispelling

KW - Feature distilling

UR - http://www.scopus.com/inward/record.url?scp=85072859390&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-29894-4_7

DO - 10.1007/978-3-030-29894-4_7

M3 - Conference contribution

AN - SCOPUS:85072859390

SN - 9783030298937

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

SP - 80

EP - 92

BT - PRICAI 2019

A2 - Nayak, Abhaya C.

A2 - Sharma, Alok

PB - Springer Verlag

T2 - 16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019

Y2 - 26 August 2019 through 30 August 2019

ER -

Bai M, Jia X, Xie W, Shen L. Adversarial Feature Distillation for Facial Expression Recognition. In Nayak AC, Sharma A, editors, PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. Springer Verlag. 2019. p. 80-92. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-29894-4_7

Adversarial Feature Distillation for Facial Expression Recognition

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this