An efficient FPGA implementation of gaussian mixture models based classifier: application to face recognition

Mehdi Neggazi, Messaoud Bengherabi, Z. Boulkenafet, Abbes Amira

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This work aims to propose an efficient hardware/software system fo guassian mixture model (GMM) parts-based topology modeling for face identification and verification. Following its great success in speaker recognition, The GMM approach was extended to face recognition providing a good trade-off in terms of complexity, performance and robustness. Despite its reduced complexity compared to other statistical modeling techniques like hiden markov model (HMM) and its variants. The GMM scoring module still to be computationally intensive algorithm consisting of a series of complex tasks executed in sequential order. This constraint limits its suitability for real-time pattern recognition embedded applications. This paper presents an efficient hardware implementation of embedded GMM based classifier. Reconfigurable system in the form of field programmable gate arrays (FPGA) is deployed to embed the hardware part of the proposed system. Furthermore a design of exponential calculation circuit is proposed for the best compromise between effectiveness and complexity. Approximations are also developed to reduce the hardware complexity. The developed system performs the identification process of an unknown input pattern over 200 models in 23 seconds, our performance evaluation indicates that a speedup of around 5.1X can be achieved over an optimized software implementation running on a 33GHz core i3 processor. A results precision of 10(-2) is obtained after performing the GMM calculation using the proposed hardware/software system.
Original languageEnglish
Title of host publication2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA)
PublisherIEEE
Pages367-371
Number of pages5
ISBN (Print)978-1-4673-5540-7
DOIs
Publication statusPublished - 2013

Keywords

  • FPGA
  • Face recognition
  • GMM
  • Reconfigurable architectures
  • Embedded systems

Cite this

Neggazi, M., Bengherabi, M., Boulkenafet, Z., & Amira, A. (2013). An efficient FPGA implementation of gaussian mixture models based classifier: application to face recognition. In 2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA) (pp. 367-371). IEEE. https://doi.org/10.1109/WoSSPA.2013.6602393
Neggazi, Mehdi ; Bengherabi, Messaoud ; Boulkenafet, Z. ; Amira, Abbes. / An efficient FPGA implementation of gaussian mixture models based classifier : application to face recognition. 2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA). IEEE, 2013. pp. 367-371
@inproceedings{6985fe908d114afcaca568c0e379907b,
title = "An efficient FPGA implementation of gaussian mixture models based classifier: application to face recognition",
abstract = "This work aims to propose an efficient hardware/software system fo guassian mixture model (GMM) parts-based topology modeling for face identification and verification. Following its great success in speaker recognition, The GMM approach was extended to face recognition providing a good trade-off in terms of complexity, performance and robustness. Despite its reduced complexity compared to other statistical modeling techniques like hiden markov model (HMM) and its variants. The GMM scoring module still to be computationally intensive algorithm consisting of a series of complex tasks executed in sequential order. This constraint limits its suitability for real-time pattern recognition embedded applications. This paper presents an efficient hardware implementation of embedded GMM based classifier. Reconfigurable system in the form of field programmable gate arrays (FPGA) is deployed to embed the hardware part of the proposed system. Furthermore a design of exponential calculation circuit is proposed for the best compromise between effectiveness and complexity. Approximations are also developed to reduce the hardware complexity. The developed system performs the identification process of an unknown input pattern over 200 models in 23 seconds, our performance evaluation indicates that a speedup of around 5.1X can be achieved over an optimized software implementation running on a 33GHz core i3 processor. A results precision of 10(-2) is obtained after performing the GMM calculation using the proposed hardware/software system.",
keywords = "FPGA, Face recognition, GMM, Reconfigurable architectures, Embedded systems",
author = "Mehdi Neggazi and Messaoud Bengherabi and Z. Boulkenafet and Abbes Amira",
year = "2013",
doi = "10.1109/WoSSPA.2013.6602393",
language = "English",
isbn = "978-1-4673-5540-7",
pages = "367--371",
booktitle = "2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA)",
publisher = "IEEE",
address = "United States",

}

Neggazi, M, Bengherabi, M, Boulkenafet, Z & Amira, A 2013, An efficient FPGA implementation of gaussian mixture models based classifier: application to face recognition. in 2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA). IEEE, pp. 367-371. https://doi.org/10.1109/WoSSPA.2013.6602393

An efficient FPGA implementation of gaussian mixture models based classifier : application to face recognition. / Neggazi, Mehdi; Bengherabi, Messaoud; Boulkenafet, Z.; Amira, Abbes.

2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA). IEEE, 2013. p. 367-371.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - An efficient FPGA implementation of gaussian mixture models based classifier

T2 - application to face recognition

AU - Neggazi, Mehdi

AU - Bengherabi, Messaoud

AU - Boulkenafet, Z.

AU - Amira, Abbes

PY - 2013

Y1 - 2013

N2 - This work aims to propose an efficient hardware/software system fo guassian mixture model (GMM) parts-based topology modeling for face identification and verification. Following its great success in speaker recognition, The GMM approach was extended to face recognition providing a good trade-off in terms of complexity, performance and robustness. Despite its reduced complexity compared to other statistical modeling techniques like hiden markov model (HMM) and its variants. The GMM scoring module still to be computationally intensive algorithm consisting of a series of complex tasks executed in sequential order. This constraint limits its suitability for real-time pattern recognition embedded applications. This paper presents an efficient hardware implementation of embedded GMM based classifier. Reconfigurable system in the form of field programmable gate arrays (FPGA) is deployed to embed the hardware part of the proposed system. Furthermore a design of exponential calculation circuit is proposed for the best compromise between effectiveness and complexity. Approximations are also developed to reduce the hardware complexity. The developed system performs the identification process of an unknown input pattern over 200 models in 23 seconds, our performance evaluation indicates that a speedup of around 5.1X can be achieved over an optimized software implementation running on a 33GHz core i3 processor. A results precision of 10(-2) is obtained after performing the GMM calculation using the proposed hardware/software system.

AB - This work aims to propose an efficient hardware/software system fo guassian mixture model (GMM) parts-based topology modeling for face identification and verification. Following its great success in speaker recognition, The GMM approach was extended to face recognition providing a good trade-off in terms of complexity, performance and robustness. Despite its reduced complexity compared to other statistical modeling techniques like hiden markov model (HMM) and its variants. The GMM scoring module still to be computationally intensive algorithm consisting of a series of complex tasks executed in sequential order. This constraint limits its suitability for real-time pattern recognition embedded applications. This paper presents an efficient hardware implementation of embedded GMM based classifier. Reconfigurable system in the form of field programmable gate arrays (FPGA) is deployed to embed the hardware part of the proposed system. Furthermore a design of exponential calculation circuit is proposed for the best compromise between effectiveness and complexity. Approximations are also developed to reduce the hardware complexity. The developed system performs the identification process of an unknown input pattern over 200 models in 23 seconds, our performance evaluation indicates that a speedup of around 5.1X can be achieved over an optimized software implementation running on a 33GHz core i3 processor. A results precision of 10(-2) is obtained after performing the GMM calculation using the proposed hardware/software system.

KW - FPGA

KW - Face recognition

KW - GMM

KW - Reconfigurable architectures

KW - Embedded systems

U2 - 10.1109/WoSSPA.2013.6602393

DO - 10.1109/WoSSPA.2013.6602393

M3 - Conference contribution

SN - 978-1-4673-5540-7

SP - 367

EP - 371

BT - 2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA)

PB - IEEE

ER -

Neggazi M, Bengherabi M, Boulkenafet Z, Amira A. An efficient FPGA implementation of gaussian mixture models based classifier: application to face recognition. In 2013 8TH INTERNATIONAL WORKSHOP ON SYSTEMS, SIGNAL PROCESSING AND THEIR APPLICATIONS (WOSSPA). IEEE. 2013. p. 367-371 https://doi.org/10.1109/WoSSPA.2013.6602393