Fever Tablet Facial Recognition Misidentifies Teenager, Alleges Racial Profiling

By Zach Segal, Published Aug 04, 2021, 09:37am EDT

A Michigan skating rink deployed a fever tablet with facial recognition that misidentified a Black teenager.

IPVM Image

The teenager's mother alleged racial profiling, while the skating rink owner told IPVM he purchased the unit based on a recommendation from a roller skating rink association.

Inside this note, we examine the case and the broader issues involved with facial recognition errors.

Mistaken ************** ********

IPVM Image

*** ********, ***** ********, was ****** *** ** a ******** ****** ******* rink,********* *****, ** **** ****, after ***** ************* ** face ***********, *** ***************. *** *** ******* off ** *** **** by *** ******* **** Saturday ***** *** **** she *** ****** **** she ***** ** *****. The ****’* **** *********** system ******* ********** ***** as * ********* **** who *** **** ** a ***** ** *** location ******* ** *** year, *** ***** *** never **** ** *** rink ******. *** ***** match **** *** ****** shown **** ** ****.

Fever ****** ****** ***********

*** ****’* *****, ****** LaMay, **** **** ** was ***** * ***** tablet *** ****** *********** but ** *** *** know ***** ***** *** declined ** ******* **** information.

**** ********** ***** **** ***** devices *** ********* ********** at ********* **** ***********, skewing *******, *** ************ ************** *** ******** *** human **** *********** ***********.

**** *** *** **** a **** *********** ******** of *********** *******, *** due ** *** ***-****, and ***-******* *******, **** would ******* ***-***** **** relying ** *** ****** recognition ** ******* *********.

*** ***** **** *** Detroit ******* *** ** of **** ************ *** *******:

IPVM Image

**** ** ** *** manufactured ** ***, ***** the ******* ** *** relabellers, ** ***** ** difficult ** ******** *** actually *** **** **** to *** ******* ****.

**** ***** ****** ***** are *** ***** *** facial *********** ********** *****-** or **** ***** ** for * ********'* *** employees, **** *** ********, which *********** ** **** mistakes *** ***** * more *********** ********* *******

Owner ***********

*****, *** **** *****, told **** **** *** terminal *** ********* *** fever ********* ******** ******* ****(***) **************, *** **** for **** *********** *** to *** ****** ** fights ** *** ********. He **** *** ****** was “****** ****** ********” in *** **** *** that ** ********* * 97% ********** ** * match **** ***** ******** was *******.

**** *** ****** ** find *** ******* ** a ***** ********* ******** on ****** ****. ***** ******** ** specify *** **************, *** RSA ******* ** ****** questions ***** *** ******** or ***** ******** ******** to ****** *****.

Family **** *****

*****’* ******, ****** ********, told *** ******* *** felt **** *** ****** profiling:

** **, **'* ********* racial ********* [...] ***'** just ****** ***** ***** Black, ***** **** **** glasses **** *** ******* and ****'* *** *****.

*************** *** ****** ** considering ****** ***** ******.

*** **** **** *** family *** ** ********* they *** **** *** Detroit:

Facial *********** ******** ***********

***** ****** *********** ******* are * ******* ***** and *******. ****** **** ***** ******** on***** ***** ** ***** arrests ***** ** **** recognition ** *** ** and *** ***** ***** Police ** *** ************% ** ******** ******* from ***** ********* **** recognition *********** ** **** and **** **** ***** positives.

***** ********** *** ******* 99%+ ******** ** ************* **** ** ****,****-***** ******* ** ***** systems *** ********* **** worse *** ****** ******** *** ******, face *********, ********* *****, ******** constraints, *** ******** *****.

Outlook - ******** ** ******** ** ***

**** ******** **** ******** to ******** ** *** facial *********** **** ******** as ******* ******** *** fundamental ** ****** ***********. Large *********** *** ******* imperfections (*******, ********, ******, etc.) **** ***** ******* problems ** ******* *** confidently *********** ** * facial *********** ***** ** valid ** ***.

Comments (7)

******'* ** **** **** extremely **** *** *** roller **** ********** ** ask *** *** ** then **** ***** **** know *** ****'* *** same ****** ** *** system **** *** ***? These ******* *** ***** going ** ** ***% accurate. * ***** *** policy ** ****** ***** facial *********** *** **** want ** *** *********** is ** *** *** ID **** ***** **** match * ******* ****** taking *** ******.

Agree: 3
Disagree
Informative: 1
Unhelpful
Funny

** *** ** * minor, *** *** *** have * ***** ****** I.D.

Agree: 3
Disagree
Informative
Unhelpful
Funny

****'* **** *** *** still **** ** *** questions ******* ** **** assuming. *** ******* ***** have **** **** *** shown ***** **. **** shows *** ************* ** a ********** **** ** not *******.

Agree: 3
Disagree
Informative
Unhelpful
Funny

*'* *** **** **'* not **** *** ************* it ** **** **'* hard ** ****** **** facial ***********. ** **** point, ****** *** ** is ****** *** **** way *** **** ** they *** *** **** an ** / **** for *** ****** ** the ********* ** **** teenager *** *** **** an **, ***.? ** either *** *** ***** a ***** **, ***.

** ***** **** ***** analytic ***** ******, ** is **** ** *** if * ***** ***** is * ******, ****, head *****, ***. **** people, **'* *** ** easy *** ** *** become ********* ******* **** truthful *** ***** ****** match *** ****** ** be ***** ***** *** match.

Agree
Disagree
Informative
Unhelpful
Funny

*** ***** ****** **** rush **** ****** *** lawsuit **** ****...

Agree: 2
Disagree
Informative
Unhelpful
Funny: 3

******* *** ******* *******, this ** * ********** fail ** ****.

******* *** *** ** was ********** ******** **** is ********* *** **** case ********.

**** *** *** ******** a *** **** ** a ***** ** * surveillance **** **** ******** of *****.

**** *** ******** * photo ***** ** *** system **** ********** ********, angle, ***, ***** ******* to * ***** *** taken **** ********** *** same ***************.

*** ** * ******** of ** **** * few ******** *****.

Agree
Disagree
Informative
Unhelpful
Funny

** ******* ******** **** this ******* ** **** roller ******* ***** *** only ***** *****, *** that **** *** ******* enough ** ******** **** a ********** ****** ******* association *** **** ** expo ** *******! *** would **** *****? * just **** **** ***'* clear *** ***** **** too **** ******* **** skates.

Agree
Disagree
Informative
Unhelpful
Funny: 2
Read this IPVM report for free.

This article is part of IPVM's 7,270 reports and 968 tests and is only available to subscribers. To get a one-time preview of our work, enter your work email to access the full article.

Already a subscriber? Login here | Join now
Loading Related Reports