Verkada Video Analytics Tested 2020

By Rob Kilpatrick and Sean Patton, Published Oct 08, 2020, 12:23pm EDT

In 2019, Verkada released people and face analytics but our testing then showed significant problems. A year later, how much better are they?

IPVM Image

Verkada's second-generation cameras tout "8x more powerful with an upgraded processor", we bought 2 of them and tested them examining:

  • How well does it detect and identify people?
  • How well does facial recognition work with hats/masks/glasses?
  • What issues does it have with environmental changes (rain, light)?
  • How well does it identify male and female appearance?
  • What distance/PPF level is required for detection?
  • How well does it identify crowds?
  • How does it compare to Avigilon, Axis, Dahua, Hikvision, etc.?

See the companion report for our test results on Verkada's 2020 camera's image quality.

Executive *******

*******'* ********* **** ******** substantially, ******** ****** **** most ** *** ****** at * ***** **** than **** ** *** market (**** / ********). People ********* *** ********, with ******* ***** ********** or ****** ******, *******, it ******** ********** ****** PPF **** ******** ********** AI *******. ***** **** no ***** *************** ** vehicles ********** ** ****** or **** *****. ************, in * ******* **********, facial *********** *** ****** analytics ********* ************* ****** than ** *** ******** test.

Competitive **********

***** *******'* ********'* *********** is ***** ******** ** some ** *** ********* competitors ****** (*.*., ********, Briefcam, *****, ***.), ** is ****** **** **** is ******** ** *****, Hikvision, *** **** ** the ******. **** **** not **** ***** ********* directly, ********* *** ***** add ***.

** ****, ** ******* the ** ****, ** configuration, *****-** ***** ********* will ** ** ********* competitive ************** *** *******. However,*******'* ***** ******* ** still ********, ********** *** ********** when ***** ********* *** enabled (** **** *** by *******).

"Backend *************" ****** / ******* ********

** *** ******* *******, Verkada *** **** ***** alerting ****** (******* ** lighting, ***** ****) *** after ******* **** **** fixed ********* '******* *************', this ***** ******** *** eliminated.

******* **** *** ****** we ***** *** ****** by "*** ** *** backend ************** *********** *** recently" *** **** ** had **** *******.

***** ** ** ******** that **** ***** ****, it **** ***** *** risk **** ******* *** unforeseen ******* ************* ******* could ******** ***** * user's ***** *********. ****** on-site ******* ** *** benefit **** ****-********, **** also ** *** **** problems ***** ********** ** they **** ****.

People *** ******* ********* **** ********

*** ********* *-****** ***** overviews *** ********* *** negatives ** *******'* *********:

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

*******'* ****** *********** ********* significantly ****** **** ******** tests, ************ ************* **** subjects **********:

IPVM Image

** ******** *****, ******** were ********** ***** ** the ******** ** ******** unique ***** ******* ** being ****** **** ***. For *******, ** *** image *****, *** **** person ** ***** ** 10 ****** ******** (*********** in ***).

**** **** ***** ********** do *** ****** *** same ***** *******. ***** clips **** ********* ***** are ***** *** ****.

IPVM Image

***** ** ***** **** duplicate ****** ********* *************, Verkada *** **** ****** a ******* ** ***** duplicates **** * ****** person:

IPVM Image

People ********* ***** / ****** *** **** ******

******* ******** ~** *** for ****** *********, ***** is ***** ** ****, reducing ******** / **** the ********* *****, ****** high ********** *********** ** have ******:

IPVM Image

***** ******* ****** (*.*.******** **,***** ***/***, ******** *********+) ******** ****** ** ~10 ***, ** *** below ************* ******** **~************* ***** * ~***.*° FoV:

IPVM Image

False ****** ********* **** ***** ******* *** ***** ****

****** ********* *** ********* consistently **** ***** ***** changes ***** ** ********* we *** *** *** when ** ****** *** first ********** *******:

IPVM Image

***** ******* ******* *** backend *************, *** ****** no ****** ********* ***** positive ****** ********* ** major ***** *******.

** **** *** **** to ****** ***** ****** detection *** ** ***** rain ***** ******* ******* its *************, *** ******* testing ****** ******** ***** alerts:

IPVM Image

Facial *********** ***** / ****** **** ***** + ****/*******

******* **** *********** *** solid, *********** ****** ******* masks, ****, **********, ** combination ********** *** *** up ** ~*' **** 25° ****-****:

IPVM Image

*******, ******* * **** with *** ********** ********, caused *** ****** ** not ****** *** ****:

IPVM Image

***** ******* ******* *** backend *************, *** ****** returned ******* **** ****** wearing ***** *** **********.

Appearance *** ****** ********* ***** / ******* ****** *******

********** ****** ********* ******** on *** *** ********** cameras *** ******** ***** and ******** *********, ** the *** ********** *** in *** ******** ****.

***** ****** ************** ********* on *** *** ********** cameras, ** * ******* population, *** *** ********** cameras ********* **** ******, without *** ***** *************** for ******* **** ** Appears ******:

IPVM Image

*******, *** ******* **** not ******** *** **** of *** ******** ******, specifically *** *******, ********** that ******* ** **** returning **** ********** ****** results.

******* *** *** ******* to ********* ********* *** improved ****** ********* ***********, we **** **** ** received.

People / ******* ***** *********

******* ****** ***** ********* performed **** ** ******* and *** ** *** to ******* **** * - ** ****** *** simultaneously ******** ** *** FoV:

IPVM Image

*******, **** ******** ********* level ****** *** **** not ******* ****** ********** violations (*.*.********,*****) **** *** ********** for ********* ****** ******.

******* ***** ********* **** performed **** ** *** parking ***:

IPVM Image

** **** *** **** to ****** ***** ***** alerts *** ** ******** after ******* ******* *** configuration, *** ******* ******* showed ******** ***** ******:

IPVM Image

Archived ***** **** **** ******* / ***** ******

******* **** ******** ** offered ** ******** (******** from *** ****** ** cloud, *** **********) *****, but *** ** **** or ******-******** *****.

** ********* **** **** non-obstructed *****, *** ***** would **** * ********* blocked ****, ******* ** unblurred ***********:

IPVM Image

***** *** *** ******* to ****** ******** ***** to **** ** ** manually **** ***** **** were ****** ** *** automatic *********.

*******

****** ** *** ******* used ** **** ****:

  • **** *** ****** **** w/30 **** *******: ~$*** USD ****** / $*** MSRP
  • ****-* ** ******* **** w/30 **** *******: ~$*,*** USD ****** / $*,*** MSRP

**** ******* ** ******* to ***** ** ******** models, **** ** ********'* H5 (~$*,***-*,***) ** ***** starlight ***** (~$*,***), ****** PNV-A9081R (~$*,*** ***), *******, all ******* ******* ******* a ~$*** ****** ************.

Versions ****

*** ******* ******* **** using *** **** "**-**-****" firmware *** *******:

IPVM Image

Comments (3)

* ***** **** ** know **** ********** ********* (both *** ******** *** maybe *** ******** ************'* version) *** **** *********** ******* *** ********** *** *** ****. Do *** **** * survey, ** ***** *** be **** ** *** together * **** ** this? **** ************* *** really ***** ***** ****** and ******** **** ***** key ********* ***** **** to ******* ********* **** gets ******* ********.

**** ** **** **-** out-of-the-box ******** *** *** do *** **** **?

Agree
Disagree
Informative
Unhelpful
Funny

*** **** *** ******** install *** *** ******* use *** ****

***:******** ***** ******** ************* 2020

Agree
Disagree
Informative
Unhelpful
Funny

**** **** ****** *** vehicle ********* *** **** our #* ******* *** a ***** ***

Agree
Disagree
Informative: 1
Unhelpful
Funny
Read this IPVM report for free.

This article is part of IPVM's 7,008 reports, 931 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