Briefcam 6.0 Tested

By Rob Kilpatrick, Updated Jun 14, 2021, 09:00am EDT

Briefcam has released version 6.0, including an overhaul of their facial recognition which claims significant performance improvements, along with mask and distance detection, as well as adding appearance search.

IPVM Image

To see how these new features and improvements really work, we tested Briefcam 6.0 for 2+ weeks, and in this 18-page report, we examine:

  • Does the 6.0 version really improve face rec? If so, by how much?
  • How, if any, has harsh angle recognition improved?
  • How, if any, has low light and IR recognition improved?
  • How, if any, has recogntion with hats and sunglasses improved?
  • What PPF is required for consistent recognition?
  • When does Briefcam produce false positives?
  • How well does mask detection work and can it be spoofed?
  • Can proximity detection accurately detect distance between people?
  • How accurate is similar appearance search?

Executive *******

***** ** *** *******, Briefcam’s *.* ******* ******** facial *********** ****** *** board, **** ** ***** recognitions ** *** ***** of ******* *** ******** of *********** ****, ****** performance ** ***** ******, in *** *****, *** in *** ******, **** significant ************ **** ******** wear ****, *******, ** masks, *** ********* ******* were ** ****** ******** as ***** ** ******, a *********** ******** ** other *********.

*******, ***** **** *** was ******, ********'* *****-******* analytics ****** ****** *** their ******* ********** ****** could *** ****** ******** across ******** *******.

*******, ********'* **** ******* ($250-500 ****), *** ******* minimum *********, *** **** hardware ************ (********* *** GPUs ** *** ****-**** and ******** *********) **** Briefcam ********* ******** ** many ***** ********* *********.

Compared ** *********

******** ** ******* ****, ******** *.* *********** Anyvision, **** ****** *********** in *** ***** *** WDR *** *** ******* performance **** ****, *****, and ******* ******, ***** in ***** ******** ********** lagged.

************, **** **** ********* does *** ******* *** video ******** ** ******* object ******* ** ********, such ** ********, *****, direction, ***, ***** **** Briefcam **** ********* ****** for ******** ******.

**** **** ** **** to ******* ** ******* test ** ********* ** the **** *** ****** and **** *********** ****** the **********.

Compared ** ********

******** ******** ******** ****** than ********'*********** ************ ***********, **** ****** confidence *********** **** **** and ***** *** ****** performance ** *** ***** and ***.

*******, ********'*********** ****** *** **** facial ********* **** ** ********/ appearance *** *** **** to **** ******** ****** changes ** *******, ***** Briefcam's ******* ********** ****** did *** ******* ********* a ******** **** ******.

************, ********** ****** ** built-in ** ******** ******* Center ********** ******** *** Appearance ****** **** *** pricing ** ***** **** Briefcam, ** $*** *** MSRP *** *******, **** no ******* ******* *****. However, ** ** ********** only **** ******** ***/*** cameras ** *****-***** ******* connected ***********'* ** *********, ***** ******** ******** supports *****-***** ******* ***/** VMS.

No ***** *********

***** * ******** ** over *** ****** ** over *** ***** ** testing ***** **** ******* and ****** *******, ** subject *** ******** *** another ****** ** *** watch ****.

IPVM Image

*******, ***** ********** *** relatively ***** ******** ** most ************ ******** ** likely ** ** ***********, such ** ******* **** several ******* ********, **** staff *** *******, ******* with ********* ** ********, and **** ******* **** thousands ** ****** ********. Results ** ****** *********** may ****, **** ***** recognition ****** **********.

Improved *********** ** ***** ******

** *** ******** *****, Briefcam ***** *** ********* subjects ******** *** ***** of **** ** ~**° angles. *******, ******* *.* accurately ********** ***** ******** at ***** ********* (** to ~**°) ***** ******* confidence ********.

IPVM Image

************, ******** *.* ************ recognized ******** ** ***** downtilts *** ****** ***** default ********** (**° *****/**° downtilt ***** *****), ********** not ********, ********** ** confidence ********.

IPVM Image

Improved *********** ** *** *****

******** *********** ** *** light (~***) **** ******** over *** **** ****, with *********** ******** ***** default "******" ********** ******** at ** ** ~**° angles. **********, *** ********** threshold ****** ** ** decreased ** *** ****** setting ** ********* ******** in ***** ******.

IPVM Image

No *********** ** ****/** (~*.****)

*******, *********** ** **** scenes **** ** ** did *** ******* * dark ******* *** *** improve **** *** **** test. ******** ***** ** recognized, *** **** **** using *** ****** ********** setting ("*****") *** ******** approaching *** ****** **** <33° ********.

IPVM Image

Improved *********** **** *****

****** ******** **** ****** through *** ***** **** identified ******* *** ***** positives ****** ********** ******* significantly, **** *********** ******** using *** ****** ********** at *** ********* (**°) and *** ***** ********** at ******* ********* (**°).

IPVM Image

Higher ********** ************ **** ********** ** ****

******** ******* ********** **** identified **** ****** ********** than **** *****., ********** only ************ ************ **** decreasing ********** ******** **** default. ***, ******** *** be ********** ***** ******* "Normal" **********.

IPVM Image

*** **** *** **** of ******** ******* ****, with ******* ***********.

IPVM Image

*******, **** ******** **** both * *** *** sunglasses, *********** *** **** accurate **** **** *****, consistently ********** **** ***** the ******* "******" ********** setting.

IPVM Image

No ********* ******* ******** ** *****

** ********* ******* **** detected ** ***** ** our *******, * ******** in **** ******** ***** and ** ***** ****** recognition ********* **** ** Anyvision *** *****. ***** below, **** **** ********* low ******* **** ******, all ******* *** ****** human *****.

IPVM Image

75-80 *** ******** *** ******** ***********

*** ******** *** ****** recognition *** *** ******, with ******** ***** ********* ~75-80 *** *** ******** facial *********** ** * full ***** *******, ******** lower **** *********** **** as ********* ** ******** (both ~** ***).

IPVM Image

Appearance ********** ****** ********

********'* ********** ********** ******, new ***** *** **** test, ********** ********** ****** similar ** * ****** of ******** ***** ** clothing, ****, ******, ***., finding *** ********* **** a ****** ******* *** same ********.

IPVM Image

Appearance ********** **** *** **** ****** ******** ******

*******, **** ****** **** not ******** ****** ******** changes. ** * ******* changes ***** ******** ** simply **** ** * jacket, **** **** *** be *****.

IPVM Image

********'* ********** ** **** is ** ****** * facial *********** ****** **** from *** ********** ******. However, **** **** ***** if *** *******'* **** was ******* **** ****** and ***** ******** *******.

IPVM Image

** ********,********'* ****** ********** ******************* **** ****** ******* as **** ** ******* appearance, ******* ******** ****** multiple **** ******* ********* a **** ******.

******** ********* **** **** was ******** ********:

**** ** *******, *** DNN **** *** **** developed ** **** ****** who **** ******* ** a ******** ****** ** interest *** **** * high ********* ** ***** the **** ****** ** based ** *** ****** wearing *** **** ******* through *** *****/*******. ** find *** **** ****** with ********* *******, *** would *** ********** **********, then *** **** *********** on *** ******* ** find **** ****** ********* days/outfits.

Appearance ********** ***** ******* *******

*************, ********'* ********** ********** also ***** ** *******. For *******, ********* *** a *** **** ****** the ****** **** *** office ***** *** ********* of **** *** ****** a ******** *********.

IPVM Image

********** ****** **** ****** all ********* ** ********* seen ** *** ******. However, ** ****** ** certain **** **** *** the **** ********.

IPVM Image

Mask ********* ******** *** ******* ** **** **** *****

********'* **** ********* (*** since *** ******** ****) accurately ******** **** ******** were ******* ***** ** the *****, **** ** subjects *** ******* ***** or ***** **** ********* shown ** ******* ****.

IPVM Image

*******, ** ***** **** be ******* ** * subject ******* ***** **** or *** ****** *** lower ****.

IPVM Image

**** ** * ****** issue ** **** ********* analytics. *** *** *******:**** ********* ********,******** **** **** ********* Tested,****** **** **** ********* Tested.

Proximity (****** **********) ********* **********

******** **** *** ******** proximity *********, ***** ******* when * ****** *** another ****** *** *** close/far, *.*., *** ****** distancing ********. *******, ** found *********** ********** **** subjects **' ***** ******** as *** ***** ******* using * *' *********.

IPVM Image

High ******** ************

** ** ***** ** our***** ****, ******** *** **** hardware ************, ********* * 4.5GHz+ *** *** **** to *** *** **** suite ** ********, ***** forensic ("** ******") *** live ********* ("**** ****") cannot ** *** ** a ****** ***.

***** *****, **** **** using * ****-******* ***** T4 ***, ** **** be ********* ** *** type ** ******** ** the *****.

IPVM Image

**** *********** ** *** common, **** ***** ********* running *** ***** ********* on * ****** ***.

*******

******** ***** ** ******** only *** ~$***-*** *** per *******, **** * 100 ******* ******* ***** (~$25,000 **** ****). ************, at ***** *** **** are ******** (*** *** real **** ******, ******* for ******** **********), ****** ~$2,000+ ** ********.

*** ******* *** (~$**,*** USD **** ****) ** available *** ******* ******* counts, ***** ******** ****** hardware *** *********.

Hardware *** ******** ****

*** ********* ******** **** used ****** *******.

  • ******** ******:***.*****.*.*

******** *** ********* ** our *** **** ****** (HPE ***** *****) **** the ********* *****:

  • **** ****** ****
  • **** ***
  • ******* ****** **** ********

******** ********* *** ****** with ** ********, * process **** **** ~** minutes.

IPVM Image

Comments (9)

"*******, ** ****** ** certain **** **** *** the **** ********."

**** ******, ****** * would ****** ****** **** to ** *** ***** mile *** ******** *** squirrel *******.

** *** ***********, ** animal ********* *** ************** an ******** ******** ** Briefcam? ** ** ***** 1 ** **** ****** classes ** ****** *** potential ***** ******?

Agree: 1
Disagree
Informative
Unhelpful
Funny: 8

** *** ********** ** me, ** *****! ** first * ******* ** was **** ***** ** the ********, *** **** we ******** ** ******** found *** **** *** across ******** ****. **'** asked **** ***** ** and **** *** ***** feedback ** *** ******. It **** ** **** in *******.

Agree
Disagree
Informative: 5
Unhelpful
Funny

**** ****** *** **** comparison. ******* ******* ** updated ******* ** *** competitors.

Agree
Disagree
Informative
Unhelpful
Funny

*** *** ******* * little **** *** ********** between "** ******" *** "Real ****"? ** **** both **** ***** ***** to ****** ** ** their * *********** **********?

Agree
Disagree
Informative
Unhelpful
Funny

** ****** ** ******** that ** ********* ******** video, ******* ****'* * face ****** ** ***** synopsis, ***** ** **** of **** ******** ** best ***** ***. ** you **** ** *** it **** *** **************, that ***** ** ** Demand.

**** **** ** ******** that ***** ** ******* (or ********** *****) *****. So ** *** **** to *** ** ***** that ******* ** * watchlist *** ****, ****'* Real ****.

Agree
Disagree
Informative: 1
Unhelpful
Funny

*** **** **** ******* you *** *** ** one ** ***?

***, * ** *** servers ** ** ***** $10K *** *** $**

Agree
Disagree
Informative
Unhelpful
Funny

*** ** ***'* ********, it's **** ** ***. We ****** ** ************ to ** ********* ** we ******'* **** ** keep *********.

***** ************** ** ****** ****, ***** ** ~$*,*** online.

*** *** *** **** streams ** * ** than ** **** ******** for ** **** ****. We **** *** ******** for ******* ** ****.

Agree
Disagree
Informative
Unhelpful
Funny

***** ***.

** ***** ** *********** to **** *** ****** cost *** ******, ********** for ****-****. *******: *** many ****-**** **** ******* you *** *** ** a ******* * ** GPU ****** **** **** ~$10K.

Agree
Disagree
Informative
Unhelpful
Funny

*********, ******, ***** **** allow ** ** **** options *** *** ********

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

This article is part of IPVM's 7,108 reports and 942 tests and is only available to members. To get a one-time preview of our work, enter your work email to access the full article.

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