Dahua, Hikvision, ZKTeco Face Mask Detection Shootout

By Zach Segal, Published Jun 19, 2020, 09:40am EDT

Temperature tablets with face mask detection are one of the hottest trends in the current coronavirus driven market. But how well does face mask detection work?

IPVM Image

We bought and tested tablet / stations from Dahua, Hikvision, and ZKTeco and shot them out side by side to see how they performed and where they failed.

Interestingly, we found 1 clear winner while the other two were very simple to fool. Inside this report, we look at the state of "AI" within these new security devices, showing where they worked and failed.

Executive ******* - ********* ****

*** ***** ********* ***** be ******* **** * dollar ****, ****, *** failed ** ****** ********** worn *****, *** *** Hikvision ********* ********** ****** than *** ***** ***. In ********, *** ******* to **** *** * mask ***** ***** ***** doubt ** *** ********** of ***** **** *********.

*** ***** ***** ********* the *******:

*** ***** ***** ********* the *********** ****** *** options:

IPVM Image

**** ********** ** *** breakdown ** ****** *****:

  • *** ***** ***** ** fooled **** * ****, but *** ********* *** not ************ ****** *** required **** *********** **** placement
  • * ****** **** **** horizontally ***** **** *** three
  • ******* * ***** ** front ** **** **** or ******** **** * cup ** ****** ****** the ***** *** ****** but *** *** ********* cameras
  • *** ******* ********** **** face *****, ***** *** nose *** *******, ** face *****
  • ** **** **** **** to **** *** ***** cameras **** * ****** strip ** ***** ******* tape *** *** ******, however *** ********* ******** much **** **** **** the ***** ***.

Fooling **** **** *********

*******, *** ***** **** detectors *** ****** ** spoofed, **** *** ***** and ****** **** ********** to **** ** ********* as ******** ******, *******, or ** *** **** of ******, ****** * beard. *** ** *** ease ** ********, ** best, ***** ******* ***** be **** ** ********* to **** * ****.

Continuous ********* ***********

** ********, ** ***** mask ******** *** ****** someone ****** ***** **** off, ** ***** ******* detection ***** **** ****** someone *** * ****, it *** ***** ****** they **** ** **. C********* **** ********* ** potentially * **** **** useful *** ** ****** people **** *****. ** will ** ******* ******* from * ******* ** vendors ** *** **** future.

************

****, *** ******** *** not ****** ****** ********** wearing *****, ** ***** below:

IPVM Image

Dahua **** **** *********

*** ***** **** **** Detection *** **** ** spoof. ******** ******* **** a *****, ****** ****, mug, *** **** ***'* hand ************ ****** *** device.

IPVM Image

*****'* ********* ******* ** focus ** *** **** right ***** *** ***** and *** ****, ** can ** **** ** the ******* ***** ***** it *** ****** ** a ***** ***** ** tape. *******, **** ***** area ** ***** *******, that ****** *** ******, a ***** *** *** counted ** * ****. However, **** ***** *** to **** ***** ************ counting ** *****.

IPVM Image

Hikvision **** **** *********

*** ********* **** ******** was **** ****** ** spoof, *** **** ********, it ****** **** ** consistently ** ** **** a ****** **** **, more ************, **** ********* ear ** ***. ** addition, **** ****** ********, it ***** ** ****** with * **** *** occasionally **** * ***.

IPVM Image

*** ******* ***** **** some **** ** ********* for ******, ** **** as *** ****, *****, and ***** *** ****. This ***** ******* *** it *** ****** ** thin **** ********* *** to ***, *** *** a *****, ** ***** below:

IPVM Image

*******, *** ********* ******** the **** *****-******** ** the *****, ******* ** recognize * ********* **** covered ***'* **** *** nose ** * ******** mask.

IPVM Image

ZKTeco **** **** *********

*** ******'* *********** *** comparable ** *** ***** model *** ******** ** focus ** * ****** area. ** *** ****** spoofed ** *** **** objects ** *** *****: a *****, ****** ****, mug, *** * ****.

IPVM Image

*******, **** **** *** fingers **** ****** ** fool **. ***, ******* because ** *** ***** on * ****** ****, simply ****** * ***** fooled ** ** ****, as ***** *****:

IPVM Image

Upcoming *******

** *** ********* *** testing ** **** **** detection **** *** **** month, ********* ******* **** Avigilon, *****, ******, *** others. ** *** **** suggestions ** **** ******* to ****, ****** ***** them ** *** ********.

Comments (10)

**** ******** *** ******* ** IPVM! **** ** ****'* first ****, ****** **** testing, ********, *** ********* from ****. ** **** joined **, ********** **** Dartmouth **** ****.

***** **** **** ***** evaluating *** ******* **** for * ** ** 20 ****** **** ** pass, *** ***** *** or *** ******** ** the ****** ** *** line **** * "**** fever".

** **** ***

******, ******. ** *** in *** ******* ** procuring * ****** ** other ******* ********* ***, hopefully **** ******* ** them ** *** **** month.

**** ***** ******* *******, I ****, **** * mask **, *** ******* on. *****'* ****** *********** for *** ****** ** match * **** *** make * **** ***********.

******, ****, ***. *** well **** ******* ******* on ***** *********?

***** *** ****, **** good ****.

**** *******.

*** ***** ** **** the ********* ******. ** are ****** **** ****** in *** ******.

** **** **** ******** more *****, **** ******* to ** * **** relabel. ** **** **** carefully ****** ** *** through *** ******* *** order *** **** ******** marketed *****.

* **** ******* *** to ********* *** **** information.

* **** *** * news ***** **** ****** City **** ** ******* out ***** ** ** the ********* ******, ** be **** *** *********** and **** **** ********* only ** *** ***********, police, *** **** *********. I **** ** ******* with ********* *** *********** using **** **********, *** I ******* **************. ****** the **** *****, ** was ****** @ *:*** that *** ***** *** made ** *** ******. You *** ******* *** from *** **** * grabbed **** *** ***** that **** *** **** in *****, *** *** MAC ******* *** **** be ****** **** ** China. *** *******'* **-************ are ** *** *****, they **** ***** * few ************ ** **, and **** *** ** my ******** ** * building ***** * *** assure *** ** ************* is ****** *****.

********* **: ****** **** is *** ** ********* body *****, **** *********** technology ** *** ********* buildings - **** *** York

IPVM Image

IPVM Image

IPVM Image

********* **: ****** **** is *** ** ********* body *****, **** *********** technology ** *** ********* buildings - **** *** York

** ***** ** ** BEMS *****. *** ****:

** ******** ********* ** "NDAA *********" *** **** or *********** ******* ** offering ****** ** *********? The ********* ** ** surely ****** **** *** cost.

********** ** **** ***** be *****. * *** a **** ** ** own *** ** ******** no **** **** * coffee *** *** **** covering ****, *** *** detect * **** **** a ****** **** *** cell *****.

Login to read this IPVM report.
Why do I need to log in?
IPVM conducts unique testing and research funded by member's payments enabling us to offer the most independent, accurate and in-depth information.
Loading Related Reports