Finessing filter scarcity problem in face recognition via multi-fold filter convolution

Cheng Yaw Low; Andrew Beng Jin Teoh

doi:10.1117/12.2280352

Finessing filter scarcity problem in face recognition via multi-fold filter convolution

Cheng Yaw Low, Andrew Beng Jin Teoh

Department of Electrical and Electronic Engineering

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

1 Citation (Scopus)

Abstract

The deep convolutional neural networks for face recognition, from DeepFace to the recent FaceNet, demand a sufficiently large volume of filters for feature extraction, in addition to being deep. The shallow filter-bank approaches, e.g., principal component analysis network (PCANet), binarized statistical image features (BSIF), and other analogous variants, endure the filter scarcity problem that not all PCA and ICA filters available are discriminative to abstract noise-free features. This paper extends our previous work on multi-fold filter convolution (â.,3-FFC), where the pre-learned PCA and ICA filter sets are exponentially diversified by â.,3 folds to instantiate PCA, ICA, and PCA-ICA offspring. The experimental results unveil that the 2-FFC operation solves the filter scarcity state. The 2-FFC descriptors are also evidenced to be superior to that of PCANet, BSIF, and other face descriptors, in terms of rank-1 identification rate (%).

Original language	English
Title of host publication	Second International Workshop on Pattern Recognition
Editors	Guojian Chen, Xudong Jiang, Masayuki Arai
Publisher	SPIE
ISBN (Electronic)	9781510613508
DOIs	https://doi.org/10.1117/12.2280352
Publication status	Published - 2017
Event	2nd International Workshop on Pattern Recognition, IWPR 2017 - Singapore, Singapore Duration: 2017 May 1 → 2017 May 3

Publication series

Name	Proceedings of SPIE - The International Society for Optical Engineering
Volume	10443
ISSN (Print)	0277-786X
ISSN (Electronic)	1996-756X

Other

Other	2nd International Workshop on Pattern Recognition, IWPR 2017
Country	Singapore
City	Singapore
Period	17/5/1 → 17/5/3

Bibliographical note

Funding Information:
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NO. 2016R1A2B4011656).

All Science Journal Classification (ASJC) codes

Electronic, Optical and Magnetic Materials
Condensed Matter Physics
Computer Science Applications
Applied Mathematics
Electrical and Electronic Engineering

Access to Document

10.1117/12.2280352

Fingerprint Dive into the research topics of 'Finessing filter scarcity problem in face recognition via multi-fold filter convolution'. Together they form a unique fingerprint.

Cite this

@inproceedings{a9804b19aeec403181e000fda0790a7c,

title = "Finessing filter scarcity problem in face recognition via multi-fold filter convolution",

abstract = "The deep convolutional neural networks for face recognition, from DeepFace to the recent FaceNet, demand a sufficiently large volume of filters for feature extraction, in addition to being deep. The shallow filter-bank approaches, e.g., principal component analysis network (PCANet), binarized statistical image features (BSIF), and other analogous variants, endure the filter scarcity problem that not all PCA and ICA filters available are discriminative to abstract noise-free features. This paper extends our previous work on multi-fold filter convolution ({\^a}.,3-FFC), where the pre-learned PCA and ICA filter sets are exponentially diversified by {\^a}.,3 folds to instantiate PCA, ICA, and PCA-ICA offspring. The experimental results unveil that the 2-FFC operation solves the filter scarcity state. The 2-FFC descriptors are also evidenced to be superior to that of PCANet, BSIF, and other face descriptors, in terms of rank-1 identification rate (%).",

author = "Low, {Cheng Yaw} and Teoh, {Andrew Beng Jin}",

note = "Funding Information: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NO. 2016R1A2B4011656).; 2nd International Workshop on Pattern Recognition, IWPR 2017 ; Conference date: 01-05-2017 Through 03-05-2017",

year = "2017",

doi = "10.1117/12.2280352",

language = "English",

series = "Proceedings of SPIE - The International Society for Optical Engineering",

publisher = "SPIE",

editor = "Guojian Chen and Xudong Jiang and Masayuki Arai",

booktitle = "Second International Workshop on Pattern Recognition",

address = "United States",

}

Low, CY & Teoh, ABJ 2017, Finessing filter scarcity problem in face recognition via multi-fold filter convolution. in G Chen, X Jiang & M Arai (eds), Second International Workshop on Pattern Recognition., 104430G, Proceedings of SPIE - The International Society for Optical Engineering, vol. 10443, SPIE, 2nd International Workshop on Pattern Recognition, IWPR 2017, Singapore, Singapore, 17/5/1. https://doi.org/10.1117/12.2280352

Finessing filter scarcity problem in face recognition via multi-fold filter convolution. / Low, Cheng Yaw; Teoh, Andrew Beng Jin.

Second International Workshop on Pattern Recognition. ed. / Guojian Chen; Xudong Jiang; Masayuki Arai. SPIE, 2017. 104430G (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 10443).

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

TY - GEN

T1 - Finessing filter scarcity problem in face recognition via multi-fold filter convolution

AU - Low, Cheng Yaw

AU - Teoh, Andrew Beng Jin

N1 - Funding Information: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NO. 2016R1A2B4011656).

PY - 2017

Y1 - 2017

N2 - The deep convolutional neural networks for face recognition, from DeepFace to the recent FaceNet, demand a sufficiently large volume of filters for feature extraction, in addition to being deep. The shallow filter-bank approaches, e.g., principal component analysis network (PCANet), binarized statistical image features (BSIF), and other analogous variants, endure the filter scarcity problem that not all PCA and ICA filters available are discriminative to abstract noise-free features. This paper extends our previous work on multi-fold filter convolution (â.,3-FFC), where the pre-learned PCA and ICA filter sets are exponentially diversified by â.,3 folds to instantiate PCA, ICA, and PCA-ICA offspring. The experimental results unveil that the 2-FFC operation solves the filter scarcity state. The 2-FFC descriptors are also evidenced to be superior to that of PCANet, BSIF, and other face descriptors, in terms of rank-1 identification rate (%).

AB - The deep convolutional neural networks for face recognition, from DeepFace to the recent FaceNet, demand a sufficiently large volume of filters for feature extraction, in addition to being deep. The shallow filter-bank approaches, e.g., principal component analysis network (PCANet), binarized statistical image features (BSIF), and other analogous variants, endure the filter scarcity problem that not all PCA and ICA filters available are discriminative to abstract noise-free features. This paper extends our previous work on multi-fold filter convolution (â.,3-FFC), where the pre-learned PCA and ICA filter sets are exponentially diversified by â.,3 folds to instantiate PCA, ICA, and PCA-ICA offspring. The experimental results unveil that the 2-FFC operation solves the filter scarcity state. The 2-FFC descriptors are also evidenced to be superior to that of PCANet, BSIF, and other face descriptors, in terms of rank-1 identification rate (%).

UR - http://www.scopus.com/inward/record.url?scp=85028548855&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85028548855&partnerID=8YFLogxK

U2 - 10.1117/12.2280352

DO - 10.1117/12.2280352

M3 - Conference contribution

AN - SCOPUS:85028548855

T3 - Proceedings of SPIE - The International Society for Optical Engineering

BT - Second International Workshop on Pattern Recognition

A2 - Chen, Guojian

A2 - Jiang, Xudong

A2 - Arai, Masayuki

PB - SPIE

T2 - 2nd International Workshop on Pattern Recognition, IWPR 2017

Y2 - 1 May 2017 through 3 May 2017

ER -