IP Camera People Counting Shootout - Avigilon, Axis, Bosch, Dahua, Hanwha, Hikvision

By Rob Kilpatrick, Published Mar 26, 2021, 10:44am EDT

People counting has become more common in IP cameras, with many manufacturers now offering it. But how well does it work?

IPVM Image

We tested people counting on multiple manufacturers' cameras, including:

The test only included conventional, generally available, IP cameras and not specialized 3D cameras.

Inside, we examine:

  • How accurate were people counts?
  • Do the analytics miss subjects or double counts?
  • How accurate can they count multiple people close together?
  • Can they count someone moving quickly?
  • Are push/pulled objects counted as people?
  • Are shadows be counted as a person?
  • Can a light change trigger a person count?
  • How does mounting angle impact count accuracy?

Manufacturer *********** *******

***** ** *** *******, Hanwha ** *** **** People ******* ********* ****, correctly ******** ****** ******* through *** ***** ***** ignoring ******* ** ***** counting **** ** ****** lingering ** *** ******** line ** ******* ** the *****.

******** ** *** ******'* X ****** ****** ******** had ****** **** ****** walking ***** ********, ************ missing * ****** ******, while ***** *** ******* subjects ******** ***** **** they ******** ** *** counting ****. ***** ****** counting ************ ********** ****** counts ** * ****** walking ****** *** ****, with ** ***** ********** as ** ***.

*******, ********* ****** ******** (on * ***** ****** camera) ********* *** ***** with ****** **** ** counting ****** ***** ** the *****, ******** ******* as ******, ** ******** multiple ***** **** ****** lingered.

Manufacturer *********** *****

**** ******** *************' ******** for ******* ***** ** view ***** *** ******** people ********. *** ************* but ******** *** ****** (AI ******) ********* *********** cameras ** ******* ******** above *** ******** **** (i.e., * *******), ****** Avigilon ***** ********* **** positioning ** **** ** reduce ****** ****** ***** in ******* *******.

*** ******* *** ******** are ***** ** *********** field ** **** *****, as ** ***** ** offered **** ***********. ************, we ****** ******* ** these ***************, **** ******* angled ****** ** ***** at *** ******** **** from *** ****, ********* later.

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

** *** *****, ********* people ******** *** *** only ****** ******** ******** to ***** ****** ***** when ******** *** ****. This *** *** ****** often, *** ** *** occur ******** ***** ****** testing.

IPVM Image

***** ***** ****** ****** to ** *** ** the *******'* ****** ***** counted ** * ****** person. ********* **** ******* counted ****** ** *** scene **** * ****** crossed *** ******** **** alone.

IPVM Image

*** **** ** *** cameras ** *** **** properly ******* ** ********** entering/exiting, ************ ****** **********.

IPVM Image

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

****** *******, **** *** people ******* *** ****** close ******** **** **** counted **** *** **** by *** ****** * series ******.

IPVM Image

************, ******** ****** * second ******* **** *** camera *** ******* ** a ***** ***** (** recommended ** ******* *******) due ** *** ****** merging ******** **** *** object, ***** ** * single ******** ***.

IPVM Image

*******, ******** *** ***** slightly ** *** ****** person *** **** ******* allowed ******** ** ***** both ****** *********. * bounding *** ** ******* around **** ******** *****.

IPVM Image

Other ********* ******** ** ***** ********

****** ******** ********* **** as **** ** ****** AI ****** *** *** miss ****** ******* ******* together ***** ******** ** exiting *** *****.

IPVM Image

Most ******* ******* ****** ******* *******

**** ******** ********* ** the **** **** **** to ***** * ****** walking ******* ******* *** scene.

IPVM Image

**** ***** ********* ****** subjects ******** *** **** at **** *****:

IPVM Image

Other ******* ****** *******

******* ** *** ***** were ******* ** ****** tested, ***** ***** ** Bosch:

IPVM Image

Person ********* ** **** ******* ******** ***** (*****/*********)

* ****** ********* ** the ********* ********* ******** counts ** ***** ****** counting. ***** **** **** these ****** *** ** eliminated ** ***** * second **** *** ********* in *** ******, ****** many *********** *** *** have **** *********/***** ***.

IPVM Image

********* ********* *********, ***** in **** *******:

IPVM Image

***** ******* ******* ***** subjects **** ****, ***** here:

IPVM Image

**** ***** *** ****** provided ** *****, ****** on ****** ********* ** the **** **** **********.

Cart ******* ** ******

** *** *****, * person ******* * **** through *** ***** *** counted ******** ***** ** Bosch.

IPVM Image

************, ********* ********* *********, incrementing ****** ****** ** 2 **** * ****** pushed * **** ******* the *****.

IPVM Image

****** ******* **** *** person *** *** *** cart.

IPVM Image

Light ******* *******

***** ******* ********* ***** counts, ***** ******* ** the ***** *** *** cause *** ** *** cameras ** ***** * person.

IPVM Image

Axis, *****, ****** * ******, *** ********* **** ******** ** ******* ******

**** ****** ******* ** a ******* ***** ** incidence, ****, *****, *** Hanwha * ******, *** Hikvision *** ********** ****** subjects. **** **** **** angle ** *** *********** by *************.

IPVM Image

******** ***, ****** **, and ***** ***** ********* counted ****** ** ***** angles, ***** *****. **** could ***** **** ************* placed ************ ******* ** be **** *** ********, instead ** *** ******** FOV. *******, ***** ******** was ********, ************* *** not ********* **** ***** of **** *****.

IPVM Image

Consider ******* *****

**** ******* ** ******** areas, ***** ****** ** careful ** ****** **** subjects **** ***** ***** the **** ** ***** or ****. ** **** cases, ******** *** ***** and **** *********** ****, resulting ** * ****** count.

IPVM Image

** **** ***** **** may ** ********* *** by ********* ** **-******* lines, *** *** *** analytics ***** *** ****.

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

**** ****** ** *** cameras **** *** *********** to *** ******** *****, counting ******* ********** ** Axis, *****, ******'* * Series, *** *********.

IPVM Image

Versions ****

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

  • ******** *.**-***-***-**: *.**.*.**
  • **** ****** *******: ******* 3.16.2
  • ***** ***-****-*: *.**.****
  • ***** ***-*********-***: **.***.*******.*.*, ***** Date: ****-**-**
  • ********* **-*********-****: **.*.** ***** 190919
  • ****** ***-******: *.**.**
  • ****** ***-*****: *.**.**

Comments (30)

***** ****! ******* ****** thought. * ****** *** a *** **** * narrow *** ** *** integrated ** ****** ******** from ******* ** *****. Probably **** ** **** reasons ;).

***

Agree
Disagree
Informative
Unhelpful
Funny

******:

***** ** ******* ** Bosch *** ********** ******** a ****** ********* ** the **** **** ***** script ******** ** *****.

Agree: 2
Disagree
Informative: 1
Unhelpful
Funny

* ***** **** ******* this *** * ****** unit ** **** **** Dahua (******** ****) ***** is ******* ***** *** likely ***** **** ***** than ****, ***** *** Avigilon

***** ****** ******** ****** 3D

Agree: 1
Disagree
Informative
Unhelpful
Funny

** *** ***** ******, this ** * **** of ** ****** ****** counting. **** **** *** a ** ***** -**** *****-* ** ****** Counter | **** **************** *** *** **** that ******.

** ***** ** ***** if ***** ******** ******** that ****** ******* ** telling ** ** **** other ********.

Agree: 7
Disagree: 1
Informative: 1
Unhelpful: 1
Funny: 6

********* ** **** **** it ***** ***** *** better **** ******** ****. I **** *** *** issues **** *** **** raised

Agree
Disagree: 1
Informative
Unhelpful
Funny: 1

* ******'* *** ****. Dahua ********* ****** ****** at ****** *** **** speeds, ***** ****** *** not ******** ****.

** *******, ***** ** our *******, **** *** camera ** ****** ** accurately ********* *** ******** the ****** ** *** object. *** *******, *****, instead ** * ****** bounding *** ****** *** subject, *** ****** ***** places *********, *** ******* rapidly ** ** *****. Because ** ****, *** camera *****'* ******* *** a ****** ****** ******** the ****, ********* ** a ****.

***** *** ***** ****** than *** ********* ****** in **** ****, *** more **** **** *** Hanwha's **.

Agree: 1
Disagree
Informative
Unhelpful
Funny

****,

*** **** *** *** shows **** ***** ***'* performing ** *** ******** to **** ***********.

************, ** *** **** to ** ****** ******** right, *** *** * camera ********* ******** *** it. ****'* **** ****** sense.

*** ******* ** *** wizmind ****** ** **** designed *** ********** ** have ** ********** ***** visitors ** * ***** etc.

**** *******,

******

Agree: 2
Disagree: 1
Informative: 1
Unhelpful
Funny

** *** **** ** do ****** ******** *****, you *** * ****** specially ******** *** **. That's **** ****** *****.

********* ***** ******* ****** *********, ** ** ************ designed *** ****, ****** uses '**** ******** *********':

IPVM Image

** *** **** ***** to ****** *** *********** or *** * ********** that "**** ******** *** ********** to **** ** ********** about ******** ** * store"?

Agree
Disagree
Informative: 1
Unhelpful
Funny: 7

** ****** *** *** correct ****.

***** **** ***** *** offering ****** ** ******** and *******.

*** *** ******* *** raw **** ******* ****

**** ** *** *****'* leading ********* ** ***** surveillance

** ***** ** ** better ** *** “** John ********’* ******** *******”

Agree: 1
Disagree
Informative: 1
Unhelpful: 4
Funny: 5

*** *** **** ***********-******* ***+ **** *********** *******-******** ********** *****(**** ****** ************ ** your *** ***** *******).

** *** **** ******* source **** *** ***** is **** ************* **** IPVM ** ***** ************, you *** ******* ** make **** **** **** (with *********, ***.).

Agree: 4
Disagree: 1
Informative: 1
Unhelpful: 1
Funny: 1

“*” **** ******** ********* study.

*** ***** **** ***********?

*** ** ********** **** them ******** ** ******** of ******* *** *********** as **** ** ***** governments.

********

********* ***** * “***** Tank”. **** *** *** information ****** *** ****** a ******* ** *** fashion.

** ******, **** **** over **,*** ********* ******** papers *** **** * few **********.

**, *** *** ****** the ***** *** *** “surveillence ******** ******** *********** with **** ** *********”

**, ***** * *** merely ********** ******* * different ****** ** ****** the ******** ** **** does **** *** ****** to ****** ****, *** took *** *********** ** change *** *********.

Agree: 1
Disagree
Informative: 1
Unhelpful: 1
Funny

*** ***** **** ***********?

** *** ** **** company **** *** ****** response ** *** *****? The **** ********* ******** it. *** *** ******* was ********.

*** *** ********* *** 'Rand ***********' *** ****'* a *********** ********* *** a ****** ******** ** our *****.

***** * *** ****** suggesting ******* * ********* camera ** ****** *** industry ** **** **** work *** ******

** *** ******** ***** does **** **** ** claims ** *** ********* does **** (*** *** we ******) **** *** actually ****, ** **** own *********. *** ****, #1, ** *** ** many ******* ****** ** not ***** ***** ***** products. *** **** **** is * ******* ******* with *****, **** ****** features **** **** ****** and **** ***** *********** make ******* *** **.

Agree: 5
Disagree
Informative: 1
Unhelpful
Funny: 1

******? **?

** *** ***** *** are ********** *** **** Rand *** ** ********** citation.

***** *** *** **** honesty, *** ****** **** header ** ****** **** business ******** ** *** so **** *** ******.

Agree
Disagree: 1
Informative
Unhelpful: 2
Funny

** *** ***** *** are ********** *** **** Rand *** ** ********** citation.

*** *** ******** **** is **** ** ** authority ** ***** ************ than ****? ****** *******.

** **** ** **** point, **** *** ****** provide ******** ** ****'* expertise ** ***** ************ as * *** ****** on ****.

Agree: 4
Disagree
Informative
Unhelpful
Funny

** ****** ****, *** control *** ***** ** you **** ** *****.

******* ** ** ***. No?

Agree
Disagree: 5
Informative
Unhelpful: 2
Funny

#*, *** ********** ** for ****** **** ** are *** "*****'* ******* ********* ** video ************"? * ********* confirming **** ** *** sharing ****, ****** *** for * **************. *** responded **** ****. ** I ******* ** ************* that **** ******** ** that **** *** **** authority / ********* ** video ************?

*****, ** **, * see ** ******** ******* up **** *****. ** I ************* **** ***** here, ****** ******* *** I **** ** ***** to *******.

Agree: 3
Disagree
Informative
Unhelpful
Funny

*** ***** **** *** typical ***** ** ********* rep (*** **** **** have **** ** ****** ventures) *** ** ****** brainwashed *** ****** ** tell ******** ********* ***** them ** **** ****! And * ******* *** find ******* ************...*** ***** make **** ***** ******* than ******* **** ********* a ***** **** ********** sucks **** ****** * know *** *** **** it (***** **** *** trunk ******** *** **** your ********). *** ********* the **** ******* *** not ****** *** ***** product ** ****. **** one *** ****... ***.

* ********* *** * whatsapp ******* ** ****** midnight **** ***** ***** offered * ***** *** Hikvision *** *** ** has ** ** ** far *** **** ******** and ********* ***** ******* I **** **** ** my ****. (***** ** happy ** ***** *** convo ** * **** screenshots ** ******** *****, it ******** **** ** busy *** ****** ***** a *** ***********). **** I ***** *** *** they *** ** **** number **** **** *** of ********. ** ****** is ******* ** ******** and * **** * private *******. * ****** integrator *** *** **** message *** ** ** across *** ******* ** the ***** ****. ***** explaining ******** ** * practising ****** **** * do *** **** *** interest ** ******* **** these ****** ** ****** very ***** *** * personal ****** *** ******* me ** **** ** China ****** *** *** how *** ******* ***** are ******* *******. * said ****** *** ** thanks. * ** *** want ***** *** ********* analytics ** **** ** up *** **** ** in * ************ **** because ** *** * look *** *** * pray ** (* ** not ****** *** ** people **** ********** **** very *******) *******, * had ** ***** *** chap. * **** ********* they **** ******* **** to **** ****...* **** pity ******...*** ***** **** much **** *** ****** defending * ***** **** has ** ***** *******, dropped ** ******* (*** best ***), ** ******* ban ***** ** ******** countries (*** **** *.*.), and ********** ***** ** bribe *** ******** ** South ******* (** **** a ******* ** ****), gets ****** *** ******* and ***** ** **** stonewall ***** **********.

*******, ********* *** ***** can ***** ***** *** nasty **** *** ** are ***** ** *** Axis, ********, *******, ******* and ******. ** * very ******* ****** **** told **...** *** ******* your ******* *** *******. And * ******* **** do *** **** ** the ********** ******* ******* of *** ************ *** industry *** ****** *** Hikvision *** * ***** told ** *****, **** were ***** ***** *** news ** *** ***** bans, ***** ******* ** major *** *** ********* (like ***) *** ****** everything ** ***** *** most ** ** ******* IPVM! ** ** **** happier *** ***** ******* selling *** **** *** of **** ********* ******** with *** ************ ** is **** *** *** takes ***** **** ** us. **** **** *** helpful ** ****** ******* and ***** ** *** fence. * ** *** think ***** *** **** anything **** ***** *** white *** ***** ** our ******** ** ****** the ***** ** *** amazing **** ** **** to ******* *** **********. We ******* ********* **** years *** **** *** big **** *** *** was ******* ******** **** than **** ******* **** link *** **** ** every ******* ** *** on *** ***** ****? We **** **** *** how *** *** **** the ***** ****** **** with **** *****? ** didn't **** *** ** assumed *** ****** ** IPVM ** **** **** doing *********** *** ***** businesses ** ********** *** government. **** ** *** good **** **** *** the **** ****, ******* this ********* ** ***** not ** **** ** inform *** *******, ***** and **** *** ******** best ** ********** ** our *** ***** *** our ********.

Agree: 1
Disagree: 1
Informative: 3
Unhelpful
Funny: 1

*** ***** **** **** money ******* **** ******* time ********* * ***** that ********** ***** **** anyone * **** *** has **** **

#*, ******! *** *********** thing ** **** *** Dahua *** ********* **** made * *** ** money **** ****. ***? Fever *******. *** *******. Also, ** ***** *** Hikvision **** ******** ***** to ***********, **** ***** jump (**** ****** **** they *** ***** ***).

* ********* *** * whatsapp ******* ** ****** midnight **** ***** ***** offered * ***** *** Hikvision *** *** ** has ** ** ** far *** **** ******** and ********* ***** ******* I **** **** ** my ****. (***** ** happy ** ***** *** convo

*'* **** ** *** it, ****** ***** ** at****@****.***, ******!

Agree
Disagree
Informative
Unhelpful
Funny

*-****** *** *** ************! I ** *** **** at *** ** *** share **.

*** *** ********* *** Korean ************* **** ***** you *********? ** ** just ******* *** ****** that ** *********?

****** *****!

Agree
Disagree
Informative
Unhelpful
Funny

*** ** ******* ********* from ******** *** *****.

Agree
Disagree
Informative
Unhelpful
Funny

** *** ***** ** the ******* **********, **** not *** ******* **********, does *** **** ****** stores *** *** ******** online *******, *** *** a **** ******** ****** app *** ** ******, has *** **** **** range *** ** ***** single *****, *** *** best ****** ********* (** tested ** ****), *** great ******* *******, *** I ** *** ****** any ******** ******* **** their ******** *** * could ** ***** ** that ***.

Agree
Disagree
Informative
Unhelpful
Funny

** ***** ******* *********** platform **** ****. * really *** *** ** without *** * **** longer **** *********** ******* a ******

Agree
Disagree
Informative
Unhelpful
Funny: 1

******

*********

*********

-----------------------------------

********

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

**** *****

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

Agree
Disagree: 1
Informative
Unhelpful
Funny: 2

*** *** ******* *** raw **** ******* ****IPVM ** *** *****'* ******* ********* ** ***** ************?

***** **** ******** **** be **********:

*)***** **** ** * *****’************************************” (******)

*)**** * ****** ***** have ** **** ** was *** ******, ***** by ****** ** ***** the ******, **’* ********* of ** ** ***********.

*)*********, *** ****** ***** not *** ******* ****** falsely ***** ** ** the ******, ** **** would ******** *** *********.

*)**** ****** ** ** the ******

*)** *** **** *****

Therefore, **** ** *** ******

*IPVM Image

Agree: 3
Disagree
Informative
Unhelpful
Funny: 5

********* ******* ******. **'* always **** ** *** how ***** ***** ****** counting ** *********** **** time. *** *** ** any ****** ******* **** for ******* ******** ******** on **** ** *** cameras? ** ***** ** interesting ** *** **** the ******* ***.

Agree: 3
Disagree
Informative
Unhelpful
Funny

** *** ********* *****, we ***'* ********* ******** in ***** ** * percentage, ***** ******** ** a ******** ** **********, which **** **** ****** depending ** ************. ** these *****, ** *** aiming ** ******* **** conditions *** ***** ******** so **** ***** *** estimate *** **** ***** conditions/issues **** ***** ** their *** ********.

Agree: 1
Disagree
Informative: 5
Unhelpful
Funny

** ****,

****, **** ******** *** title, ********** *** ******* in "**** **" ******* will ** ****** ** clarify *** ****************.

*********, ** ****** **** be **** ********. *** I *****, **** *** clarifying ****, ****** *********.

***'* *** *****?

Agree: 1
Disagree: 1
Informative
Unhelpful
Funny

* ***** * **** in *** ******* ****:

*** **** **** ******** conventional, ********* *********, ** cameras *** *** *********** 3D *******.

** *** *** ******* to ******* ** ******* from **** ** ***** or ******* ***. *** wanted ** ***** **** the **** ****** ********* conventional ******* ** *** how **** **.

*********, ** ****** **** be **** ********

***** *** **** **** showed **** ****** ** and **** ****** ******* were ***** ******** ** the ********** ******* ***** 2D *** ** ********* may ** *******.

**'** ********** ** **** video ********* ******* ** this ** **** ** the *******.

** ********* ** *** Dahua ******** ** **** this ** * ****** Dahua ****** ******* - "yeah ** *** **** model, ** **********, ***** but ******** *** ****** use **** ***** ** you ****** **** * number ** ****** ***** we ******** ******* ******** where **** *****." ** you ****** **** ** works, **** ** *****. If ** *** ********, accept **** *** ***'* make *******.

Agree: 3
Disagree
Informative: 1
Unhelpful
Funny

** ********* ** *** Dahua ******** ** **** this ** * ****** Dahua ****** ******* - "yeah ** *** **** model, ** **********, ***** but ******** *** ****** use **** ***** ** you ****** **** * number ** ****** ***** we ******** ******* ******** where **** *****." ** you ****** **** ** works, **** ** *****. If ** *** ********, accept **** *** ***'* make *******.

* ********** ***** **** you. ** **** * think **’* * “***** operandi” **** ***** ** say “***’* **** **** of *** **** *****, is *** ****” :)

*’* ******* ***** *********** between ** ** **. 2D ****** ******** ******** non ***** ******* **** shadows, ******** *****, *********….

Agree
Disagree
Informative
Unhelpful
Funny

* ***** ** ***** have **** ***** ** we ***** **** *** top ********** *** *** least **** *** *** on *** ***** ******** tests.Its * **** **** and ** **** ** appreciate *** **** **** went **** ****.

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

This article is part of IPVM's 6,898 reports, 921 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