Learning non-linear reconstruction models for image set classification

Munawar Hayat, Mohammed Bennamoun, Senjian An

Research output: A Conference proceeding or a Chapter in BookConference contribution

46 Citations (Scopus)
1 Downloads (Pure)

Abstract

We propose a deep learning framework for image set classification with application to face recognition. An Adaptive Deep Network Template (ADNT) is defined whose parameters are initialized by performing unsupervised pre-training in a layer-wise fashion using Gaussian Restricted Boltzmann Machines (GRBMs). The pre-initialized ADNT is then separately trained for images of each class and class-specific models are learnt. Based on the minimum reconstruction error from the learnt class-specific models, a majority voting strategy is used for classification. The proposed framework is extensively evaluated for the task of image set classification based face recognition on Honda/UCSD, CMU Mobo, YouTube Celebrities and a Kinect dataset. Our experimental results and comparisons with existing state-of-the-art methods show that the proposed method consistently achieves the best performance on all these datasets.
Original languageEnglish
Title of host publicationProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
EditorsMunawar Hayat
Place of PublicationColumbus OH
PublisherIEEE
Pages1915-1922
Number of pages8
Volume1
ISBN (Electronic)9781479951178, 9781479951178
ISBN (Print)9781479951178
DOIs
Publication statusPublished - 23 Jun 2014
Externally publishedYes
Event2014 IEEE Conference on Computer Vision and Pattern Recognition - Columbus, Columbus, United States
Duration: 23 Jun 201428 Jun 2014

Publication series

NameProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
ISSN (Print)1063-6919

Conference

Conference2014 IEEE Conference on Computer Vision and Pattern Recognition
Abbreviated titleCVPR 2014
CountryUnited States
CityColumbus
Period23/06/1428/06/14

Fingerprint

Face recognition
Deep learning

Cite this

Hayat, M., Bennamoun, M., & An, S. (2014). Learning non-linear reconstruction models for image set classification. In M. Hayat (Ed.), Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Vol. 1, pp. 1915-1922). [6909643] (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition). Columbus OH: IEEE. https://doi.org/10.1109/cvpr.2014.246
Hayat, Munawar ; Bennamoun, Mohammed ; An, Senjian. / Learning non-linear reconstruction models for image set classification. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. editor / Munawar Hayat. Vol. 1 Columbus OH : IEEE, 2014. pp. 1915-1922 (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition).
@inproceedings{1569cffb8628420ea3e55e680b2f8aa8,
title = "Learning non-linear reconstruction models for image set classification",
abstract = "We propose a deep learning framework for image set classification with application to face recognition. An Adaptive Deep Network Template (ADNT) is defined whose parameters are initialized by performing unsupervised pre-training in a layer-wise fashion using Gaussian Restricted Boltzmann Machines (GRBMs). The pre-initialized ADNT is then separately trained for images of each class and class-specific models are learnt. Based on the minimum reconstruction error from the learnt class-specific models, a majority voting strategy is used for classification. The proposed framework is extensively evaluated for the task of image set classification based face recognition on Honda/UCSD, CMU Mobo, YouTube Celebrities and a Kinect dataset. Our experimental results and comparisons with existing state-of-the-art methods show that the proposed method consistently achieves the best performance on all these datasets.",
keywords = "face recognition, pattern recognition, computer vision, Image Set Classification, Deep Learning, Face Recognition",
author = "Munawar Hayat and Mohammed Bennamoun and Senjian An",
year = "2014",
month = "6",
day = "23",
doi = "10.1109/cvpr.2014.246",
language = "English",
isbn = "9781479951178",
volume = "1",
series = "Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition",
publisher = "IEEE",
pages = "1915--1922",
editor = "Munawar Hayat",
booktitle = "Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition",

}

Hayat, M, Bennamoun, M & An, S 2014, Learning non-linear reconstruction models for image set classification. in M Hayat (ed.), Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. vol. 1, 6909643, Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, IEEE, Columbus OH, pp. 1915-1922, 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, United States, 23/06/14. https://doi.org/10.1109/cvpr.2014.246

Learning non-linear reconstruction models for image set classification. / Hayat, Munawar; Bennamoun, Mohammed; An, Senjian.

Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. ed. / Munawar Hayat. Vol. 1 Columbus OH : IEEE, 2014. p. 1915-1922 6909643 (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition).

Research output: A Conference proceeding or a Chapter in BookConference contribution

TY - GEN

T1 - Learning non-linear reconstruction models for image set classification

AU - Hayat, Munawar

AU - Bennamoun, Mohammed

AU - An, Senjian

PY - 2014/6/23

Y1 - 2014/6/23

N2 - We propose a deep learning framework for image set classification with application to face recognition. An Adaptive Deep Network Template (ADNT) is defined whose parameters are initialized by performing unsupervised pre-training in a layer-wise fashion using Gaussian Restricted Boltzmann Machines (GRBMs). The pre-initialized ADNT is then separately trained for images of each class and class-specific models are learnt. Based on the minimum reconstruction error from the learnt class-specific models, a majority voting strategy is used for classification. The proposed framework is extensively evaluated for the task of image set classification based face recognition on Honda/UCSD, CMU Mobo, YouTube Celebrities and a Kinect dataset. Our experimental results and comparisons with existing state-of-the-art methods show that the proposed method consistently achieves the best performance on all these datasets.

AB - We propose a deep learning framework for image set classification with application to face recognition. An Adaptive Deep Network Template (ADNT) is defined whose parameters are initialized by performing unsupervised pre-training in a layer-wise fashion using Gaussian Restricted Boltzmann Machines (GRBMs). The pre-initialized ADNT is then separately trained for images of each class and class-specific models are learnt. Based on the minimum reconstruction error from the learnt class-specific models, a majority voting strategy is used for classification. The proposed framework is extensively evaluated for the task of image set classification based face recognition on Honda/UCSD, CMU Mobo, YouTube Celebrities and a Kinect dataset. Our experimental results and comparisons with existing state-of-the-art methods show that the proposed method consistently achieves the best performance on all these datasets.

KW - face recognition

KW - pattern recognition

KW - computer vision

KW - Image Set Classification

KW - Deep Learning

KW - Face Recognition

UR - http://www.scopus.com/inward/record.url?scp=84908410915&partnerID=8YFLogxK

UR - http://www.mendeley.com/research/learning-nonlinear-reconstruction-models-image-set-classification

U2 - 10.1109/cvpr.2014.246

DO - 10.1109/cvpr.2014.246

M3 - Conference contribution

SN - 9781479951178

VL - 1

T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

SP - 1915

EP - 1922

BT - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

A2 - Hayat, Munawar

PB - IEEE

CY - Columbus OH

ER -

Hayat M, Bennamoun M, An S. Learning non-linear reconstruction models for image set classification. In Hayat M, editor, Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Vol. 1. Columbus OH: IEEE. 2014. p. 1915-1922. 6909643. (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition). https://doi.org/10.1109/cvpr.2014.246