Stable face representations

Rob Jenkins, A. Mike Burton

Research output: Contribution to journalArticle

96 Citations (Scopus)

Abstract

Photographs are often used to establish the identity of an individual or to verify that they are who they claim to be. Yet, recent research shows that it is surprisingly difficult to match a photo to a face. Neither humans nor machines can perform this task reliably. Although human perceivers are good at matching familiar faces, performance with unfamiliar faces is strikingly poor. The situation is no better for automatic face recognition systems. In practical settings, automatic systems have been consistently disappointing. In this review, we suggest that failure to distinguish between familiar and unfamiliar face processing has led to unrealistic expectations about face identification in applied settings. We also argue that a photograph is not necessarily a reliable indicator of facial appearance, and develop our proposal that summary statistics can provide more stable face representations. In particular, we show that image averaging stabilizes facial appearance by diluting aspects of the image that vary between snapshots of the same person. We review evidence that the resulting images can outperform photographs in both behavioural experiments and computer simulations, and outline promising directions for future research.

Original languageEnglish
Pages (from-to)1671-1683
Number of pages13
JournalPhilosophical Transactions of the Royal Society of London. Series B: Biological sciences
Volume366
Issue number1571
DOIs
Publication statusPublished - 12 Jun 2011

Keywords

  • face perception
  • face recognition
  • identification
  • security
  • familiarity
  • biometrics
  • unfamiliar faces
  • facial identification
  • eyewitness testimony
  • external features
  • matching task
  • recognition
  • familiar
  • images
  • photographs
  • prosopagnosia

Cite this

Stable face representations. / Jenkins, Rob; Burton, A. Mike.

In: Philosophical Transactions of the Royal Society of London. Series B: Biological sciences, Vol. 366, No. 1571, 12.06.2011, p. 1671-1683.

Research output: Contribution to journalArticle

Jenkins, Rob ; Burton, A. Mike. / Stable face representations. In: Philosophical Transactions of the Royal Society of London. Series B: Biological sciences. 2011 ; Vol. 366, No. 1571. pp. 1671-1683.
@article{733a4ffc7a844a4bb95c8ae581da99c8,
title = "Stable face representations",
abstract = "Photographs are often used to establish the identity of an individual or to verify that they are who they claim to be. Yet, recent research shows that it is surprisingly difficult to match a photo to a face. Neither humans nor machines can perform this task reliably. Although human perceivers are good at matching familiar faces, performance with unfamiliar faces is strikingly poor. The situation is no better for automatic face recognition systems. In practical settings, automatic systems have been consistently disappointing. In this review, we suggest that failure to distinguish between familiar and unfamiliar face processing has led to unrealistic expectations about face identification in applied settings. We also argue that a photograph is not necessarily a reliable indicator of facial appearance, and develop our proposal that summary statistics can provide more stable face representations. In particular, we show that image averaging stabilizes facial appearance by diluting aspects of the image that vary between snapshots of the same person. We review evidence that the resulting images can outperform photographs in both behavioural experiments and computer simulations, and outline promising directions for future research.",
keywords = "face perception, face recognition, identification, security, familiarity, biometrics, unfamiliar faces, facial identification, eyewitness testimony, external features, matching task, recognition, familiar, images, photographs, prosopagnosia",
author = "Rob Jenkins and Burton, {A. Mike}",
year = "2011",
month = "6",
day = "12",
doi = "10.1098/rstb.2010.0379",
language = "English",
volume = "366",
pages = "1671--1683",
journal = "Philosophical Transactions of the Royal Society of London. Series B: Biological sciences",
issn = "0264-3960",
number = "1571",

}

TY - JOUR

T1 - Stable face representations

AU - Jenkins, Rob

AU - Burton, A. Mike

PY - 2011/6/12

Y1 - 2011/6/12

N2 - Photographs are often used to establish the identity of an individual or to verify that they are who they claim to be. Yet, recent research shows that it is surprisingly difficult to match a photo to a face. Neither humans nor machines can perform this task reliably. Although human perceivers are good at matching familiar faces, performance with unfamiliar faces is strikingly poor. The situation is no better for automatic face recognition systems. In practical settings, automatic systems have been consistently disappointing. In this review, we suggest that failure to distinguish between familiar and unfamiliar face processing has led to unrealistic expectations about face identification in applied settings. We also argue that a photograph is not necessarily a reliable indicator of facial appearance, and develop our proposal that summary statistics can provide more stable face representations. In particular, we show that image averaging stabilizes facial appearance by diluting aspects of the image that vary between snapshots of the same person. We review evidence that the resulting images can outperform photographs in both behavioural experiments and computer simulations, and outline promising directions for future research.

AB - Photographs are often used to establish the identity of an individual or to verify that they are who they claim to be. Yet, recent research shows that it is surprisingly difficult to match a photo to a face. Neither humans nor machines can perform this task reliably. Although human perceivers are good at matching familiar faces, performance with unfamiliar faces is strikingly poor. The situation is no better for automatic face recognition systems. In practical settings, automatic systems have been consistently disappointing. In this review, we suggest that failure to distinguish between familiar and unfamiliar face processing has led to unrealistic expectations about face identification in applied settings. We also argue that a photograph is not necessarily a reliable indicator of facial appearance, and develop our proposal that summary statistics can provide more stable face representations. In particular, we show that image averaging stabilizes facial appearance by diluting aspects of the image that vary between snapshots of the same person. We review evidence that the resulting images can outperform photographs in both behavioural experiments and computer simulations, and outline promising directions for future research.

KW - face perception

KW - face recognition

KW - identification

KW - security

KW - familiarity

KW - biometrics

KW - unfamiliar faces

KW - facial identification

KW - eyewitness testimony

KW - external features

KW - matching task

KW - recognition

KW - familiar

KW - images

KW - photographs

KW - prosopagnosia

U2 - 10.1098/rstb.2010.0379

DO - 10.1098/rstb.2010.0379

M3 - Article

VL - 366

SP - 1671

EP - 1683

JO - Philosophical Transactions of the Royal Society of London. Series B: Biological sciences

JF - Philosophical Transactions of the Royal Society of London. Series B: Biological sciences

SN - 0264-3960

IS - 1571

ER -