Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models

Muhammad Ali Farooq; Peter Corcoran

doi:10.1109/ICCE50685.2021.9427690

Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models

Muhammad Ali Farooq
, Peter Corcoran

Electrical & Electronic Engineering

University of Galway

Research output: Chapter in Book or Conference Publication/Proceeding › Conference Publication › peer-review

4 Citations (Scopus)

Abstract

Thermal imaging has played a dynamic role in the diversified field of consumer technology applications. To build artificially intelligent thermal imaging systems, large scale thermal datasets are required for successful convergence of complex deep learning models. In this study, we have highlighted various techniques for generating large scale synthetic facial thermal data using both public and locally gathered datasets. It includes data augmentation, synthetic data generation using StyleGAN network, and 2D to 3D image reconstruction using deep learning architectures. Training and validation accuracy of Wide ResNet CNN for binary gender recognition task is improved by 4.6% and 4.4% using original and newly generated synthetic data with an overall test accuracy of 83.33%.

Original language	English
Title of host publication	2021 IEEE International Conference on Consumer Electronics, ICCE 2021
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781728197661
DOIs	https://doi.org/10.1109/ICCE50685.2021.9427690
Publication status	Published - 10 Jan 2021
Event	2021 IEEE International Conference on Consumer Electronics, ICCE 2021 - Las Vegas, United States Duration: 10 Jan 2021 → 12 Jan 2021

Publication series

Name	Digest of Technical Papers - IEEE International Conference on Consumer Electronics
Volume	2021-January
ISSN (Print)	0747-668X

Conference

Conference	2021 IEEE International Conference on Consumer Electronics, ICCE 2021
Country/Territory	United States
City	Las Vegas
Period	10/01/21 → 12/01/21

Keywords

Augmentation
Deep Neural Networks
GAN
Infrared Imaging
LWIR
Synthetic Data
Wide ResNet

Access to Document

10.1109/ICCE50685.2021.9427690

Cite this

Farooq, M. A., & Corcoran, P. (2021). Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models. In 2021 IEEE International Conference on Consumer Electronics, ICCE 2021 Article 9427690 (Digest of Technical Papers - IEEE International Conference on Consumer Electronics; Vol. 2021-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICCE50685.2021.9427690

Farooq, Muhammad Ali ; Corcoran, Peter. / Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models. 2021 IEEE International Conference on Consumer Electronics, ICCE 2021. Institute of Electrical and Electronics Engineers Inc., 2021. (Digest of Technical Papers - IEEE International Conference on Consumer Electronics).

@inproceedings{9d76ef7f072643d6b7ed1064d03a50cf,

title = "Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models",

abstract = "Thermal imaging has played a dynamic role in the diversified field of consumer technology applications. To build artificially intelligent thermal imaging systems, large scale thermal datasets are required for successful convergence of complex deep learning models. In this study, we have highlighted various techniques for generating large scale synthetic facial thermal data using both public and locally gathered datasets. It includes data augmentation, synthetic data generation using StyleGAN network, and 2D to 3D image reconstruction using deep learning architectures. Training and validation accuracy of Wide ResNet CNN for binary gender recognition task is improved by 4.6\% and 4.4\% using original and newly generated synthetic data with an overall test accuracy of 83.33\%.",

keywords = "Augmentation, Deep Neural Networks, GAN, Infrared Imaging, LWIR, Synthetic Data, Wide ResNet",

author = "Farooq, \{Muhammad Ali\} and Peter Corcoran",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 2021 IEEE International Conference on Consumer Electronics, ICCE 2021 ; Conference date: 10-01-2021 Through 12-01-2021",

year = "2021",

month = jan,

day = "10",

doi = "10.1109/ICCE50685.2021.9427690",

language = "English",

series = "Digest of Technical Papers - IEEE International Conference on Consumer Electronics",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "2021 IEEE International Conference on Consumer Electronics, ICCE 2021",

}

Farooq, MA & Corcoran, P 2021, Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models. in 2021 IEEE International Conference on Consumer Electronics, ICCE 2021., 9427690, Digest of Technical Papers - IEEE International Conference on Consumer Electronics, vol. 2021-January, Institute of Electrical and Electronics Engineers Inc., 2021 IEEE International Conference on Consumer Electronics, ICCE 2021, Las Vegas, United States, 10/01/21. https://doi.org/10.1109/ICCE50685.2021.9427690

Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models. / Farooq, Muhammad Ali; Corcoran, Peter.
2021 IEEE International Conference on Consumer Electronics, ICCE 2021. Institute of Electrical and Electronics Engineers Inc., 2021. 9427690 (Digest of Technical Papers - IEEE International Conference on Consumer Electronics; Vol. 2021-January).

Research output: Chapter in Book or Conference Publication/Proceeding › Conference Publication › peer-review

TY - GEN

T1 - Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models

AU - Farooq, Muhammad Ali

AU - Corcoran, Peter

PY - 2021/1/10

Y1 - 2021/1/10

N2 - Thermal imaging has played a dynamic role in the diversified field of consumer technology applications. To build artificially intelligent thermal imaging systems, large scale thermal datasets are required for successful convergence of complex deep learning models. In this study, we have highlighted various techniques for generating large scale synthetic facial thermal data using both public and locally gathered datasets. It includes data augmentation, synthetic data generation using StyleGAN network, and 2D to 3D image reconstruction using deep learning architectures. Training and validation accuracy of Wide ResNet CNN for binary gender recognition task is improved by 4.6% and 4.4% using original and newly generated synthetic data with an overall test accuracy of 83.33%.

AB - Thermal imaging has played a dynamic role in the diversified field of consumer technology applications. To build artificially intelligent thermal imaging systems, large scale thermal datasets are required for successful convergence of complex deep learning models. In this study, we have highlighted various techniques for generating large scale synthetic facial thermal data using both public and locally gathered datasets. It includes data augmentation, synthetic data generation using StyleGAN network, and 2D to 3D image reconstruction using deep learning architectures. Training and validation accuracy of Wide ResNet CNN for binary gender recognition task is improved by 4.6% and 4.4% using original and newly generated synthetic data with an overall test accuracy of 83.33%.

KW - Augmentation

KW - Deep Neural Networks

KW - GAN

KW - Infrared Imaging

KW - LWIR

KW - Synthetic Data

KW - Wide ResNet

UR - https://www.scopus.com/pages/publications/85105983116

U2 - 10.1109/ICCE50685.2021.9427690

DO - 10.1109/ICCE50685.2021.9427690

M3 - Conference Publication

AN - SCOPUS:85105983116

T3 - Digest of Technical Papers - IEEE International Conference on Consumer Electronics

BT - 2021 IEEE International Conference on Consumer Electronics, ICCE 2021

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2021 IEEE International Conference on Consumer Electronics, ICCE 2021

Y2 - 10 January 2021 through 12 January 2021

ER -

Farooq MA, Corcoran P. Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models. In 2021 IEEE International Conference on Consumer Electronics, ICCE 2021. Institute of Electrical and Electronics Engineers Inc. 2021. 9427690. (Digest of Technical Papers - IEEE International Conference on Consumer Electronics). doi: 10.1109/ICCE50685.2021.9427690

Proof-of-Concept Techniques for Generating Synthetic Thermal Facial Data for Training of Deep Learning Models

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this