Hikvision AcuSense G2 Camera Test

By Rob Kilpatrick, Published Oct 29, 2020, 12:55pm EDT

Hikvision has released their next generation of AcuSense analytic cameras claiming "events such as rain drops, animals, or the movement of foliage, for example, will be drastically disregarded."

IPVM Image

We bought and tested the Hikvision DS-2CD2346G2-I examining:

  • How does AcuSense G2 compare to Hikvision's first generation AcuSense cameras?
  • How does AcuSense G2 compare to Avigilon, Bosch, and Dahua analytics?
  • Does it ignore common indoor and outdoor sources of false alarms?
  • Were people and cars accurately detected?
  • Are people properly detected walking, running, or crawling?
  • How well does it integrate with major VMSes?
  • What kind of SoC is used?
  • How does image quality compare to the first generation Acusense?

Executive *******

** * **** ** testing, ******** ** ******* most ***** ***** *******, such ** ***** ****, shadows, ******, *** *******, with ******** ********* ** humans (********* ******* *** crawling) ** ~** ***, ranges ********** **** *** performers ** ********* ********* ********. ************, ***** ******* in *** ***** *** dark/IR ****** ******** **** the ******** ** ******.

*******, ** ******** **** occasional ***** ****** ** small/large ******* (*-* *** day) *** *** *********** was ****** **** *** G1 ******** ******, **** significant ************ ** ******** against ****** *********.

Non-Production ********

****** *******, ********* ******** us **** ******** *** yet ******** ********* (**.*.*** build ******) ** ******* animal ***** ***** ******. However, **** *** ******** introduced ***** ****** ** parked ******** (***** *****), not **** ** ********** firmware (**.*.*** ***** ******) ****** **** *** a ******* ***** ** our **** ** ******** G1.

** ********* ***** ******* improvements ** ******** ** address ***** ******, ** will ****** *** ****** this ******.

Improvements **** ******** ********

******** ** *** ******* improvements ******** ** ** in *** ******** *******, including ********* ** ******* subjects, ****** ********* ** animals, *** ******** ************** of ****** (** ********** classified ****** ** ******** in ****

  • ********* ** ******* ********:******** ** ********** ****** running ********, *** ** accurately ******** ******* ******** in *** ******.
  • ****** ********* ** *******:** ***** ******* ** animals ********** ***** *** basic *******, ********** *** target ******** ** "*******" improved *********** *** *** not ********** *** *** of *** ****** ***** still ******** ~*-* ***** per ***. ** ******** from ******** ****** ** animals ********** ** ********.
  • ****** *** ********** ** vehicles:******** ** *** *** misclassify ****** ** ********. G1 ********** ********** ****** as ******** ** **** scenes.

*******, ** *** ** difference ** ***** ***** which ********* ******** ** improved ** ***** *********, such ** ****** ***-***** detection *** ****** ********* area. ** *** ** performed *********** *** **** in *** ***** *** both ******** ~** *** for ******** ***** ******* detection.

Vs. ********, *****, *****

******** ** *********** ** close ** **** ** top ********** **** ********* ********* ********:******** ***,***** ***, ******** *********+. ***** *** ** analytics ****** **** ********** alerts ** *******, ***** issues **** *** ******** in *** *******, *** animals ***** ****** *** be ******* ** **** applications (*.*., ******* ** more ***** ** ******** areas).

******** *** ***** ******** models *** ************* **** expensive **** ******** **, selling *** ~$*** ** more **. $*** *** Acusense. ***** ****** ******* models ***** *** ****** similarly ** ******** ** cameras.

** *** **, ******** G2 ************ ** ******* to *** *** *** fixed **** ****** (****** G1 ********* ****** *** available). ** ***** *******, wider ******* ** *********, with **** *********** *** motorized **** *******.

Pricing **. ***-******** ******

*** ********* **-*********-* *** be ***** ****** *** ~$200 ***, ******** **** than *** ******** ******** line (~$*** ******).

**** ******* ** * small ** ******** ******* over ***-******** ******. *** example, * *** ********* Value ****** ****** (**-*********-*) sells *** ~$***-*** *** online, $**-** ***** **** AcuSense **. ******** ** a*** ******* ****** (~$*** online), ***** ******** **** Hikvision's **** ********** ********* analytics, ******* ** **** slightly ******.

"Target ********" *****

******** ** ********** ** additional "****** ********" ******* not ***** ** *** original ******** ****. **** setting ****** **** "*****" to "*******", ***** ** effectively * ************** ********** level.

IPVM Image

** *** *******, ** found ** ******** ** using "*******" ****** ********, with ** ****** ********* or ******* ***** ******** to ***** ********.

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

******* ******* *** **** an ***** **** ***** "Basic" ********. *******, ***** alerts **** ** ****** present **** ********** ** "Highest."

IPVM Image

False ****** ** ***** / ***** *******

**** ***** ******* ******** ("Basic"), *** ********* ********** alerted ** ***** *** large *******. ***** ****** could *** ** ********** by ***** ********* ****** Validity ********, **** *-* alerts *** *** ***** occurring ** "*******."

IPVM Image

People ********** ******** *******, ********, **********

******** ** ********** ******** people *******, ********, *** obstructed **** ** ****** of ****** ** *** scene ****** *******.

IPVM Image

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

***** ****/******* ********, ***** rain *** ******** *** not ***** *** ***** alerts ** *** *****. Objects *** ******* ** seen *********** **

******* ****** ** *** scene **** **** *** alarmed **.

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

** *** *******, ****** indoor ******* ** ***** alarms **** ** ***** light ******* **** *******.

************, ******* ****** ********** the ***** **** **** ignored ** *** ******** G2 ******.

Human *** ******* ********** ****** *********

******** ** **** **** human *** ******* ************** to ****** *********, *** found ** ***** ********* models. *******, **** ******** effectively ******* ** ********* Detection ***** "*****" ****** Validity ********, ********* ** false ****** ** ******* and ******* *******.

IPVM Image

~10 *** ****** *** ****** *********

** *** *****, ******** G2 ******** ******** ** ~180' ********, ~** *** in *** *** **-*********-* camera ** ******. ***** PPF ************ *** ***, comparable ***** ********** ********* ** our *********, **** ********** ********** ***.

IPVM Image

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

**** *********** ******** **** the **, ******** **** chart ******* ** ~** PPF.

IPVM Image

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

*** ***** ***** ******* improved, ******** ******* ***** details ** ~** ***'

IPVM Image

Slightly ****** **** ***** ***********

** *** *******, **** light *********** *** ******** better, ********** ** ***** PPF ****** (~** *** below), **** ******** ****** facial ******* *** ****** text **********.

IPVM Image

Poor *** ***********

** *** ***** *** performance *** ***** **** the ** ******** ******, overexposing *** ******* ******* of *** ***** *** drastically ******** ****** ******* against ****** *********.

IPVM Image

VMS ***********

The ********* ** ****** ******* ********** ****** **** ***********, *********, *** ** ******* *** *** *** ********* **** ******* ** ********.

IPVM Image

Unknown ***

*** ******** ** **** a *** **** ** branding, **** *****.

IPVM Image

**** ******* **** ******** G1 ***** **** ** Ambarella *** ***** **** a ******** ****** ****** Processing **** (***).

Versions ****

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

  • **-*********-*:**.*.*** ***** ******

Comments (15)

**** **** ***** **** a ***** ***?

***, *** **-*********-* **** we ****** **** *** have *** ***** ***** LED.

***,

*** ** *** **** with ***** ***

** **** **** ** option ** **** ***** LED ** **** ********* trigger

*** **** *** ** set ** ** ** sec (***** ******)

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

** *** ** *** with **, ***** ***, and *******

****'* *** ***** ****** of **** ****** *** what ****** *** *** in?

* ** ** ******

*** ***

*'* ** ****** ***. "HIK ***" *****'* **** me. ****'* ******** ******? ***'* ******* *****'* even **** ***** ******** turrets (*** ***** *** 8MP ******) *** *** only *** *** ** ADI ** *** **-*********-* with ** *****/*******. **** the *** **-*********-***/** *** those ********.

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

***** *** ******* *** NDAA, ******* ** *** a ********* ****** *** for *** ***** **** insight **** ********* *** stock ********** ** *** near ****.

*** *** **** ******** eventually *** *** ** the ***** ********* ** animals ** *** ** only ********* ***** ********* on ****** ****?

***** *** *** ************* phrases * ***** ** the *******. * **** wanted ** ******* ** the ****** ***** ****** are ****** **** *** newer **** ********.

- "** *** *** false ***** ** ******* when ***** ****** ********** settings, ****** **** ****** were ***** ******** ***** default ********."

- "**** ***** ******* settings ("*****"), *** ********* frequently ******* ** ***** and ***** *******. ***** alerts ***** *** ** eliminated ** ***** ********* Target ******** ********, **** 1-2 ****** *** *** still ********* ** *******."

*** **** ******** ******* the ****** ** ***** alarms ** ******* *** still *** *** *** rid ** **** **********, there **** ***** *-* a *** ***** *** "Highest" ****** ******** *******. Thanks *** ******** **** out, * ********* ** in *** ******.

** *** ***, **** IPVM ***** **** *** made **** *** ********** or *** **** ****** that ** ** ********* else?

** *** **** *** SoC **********, ***** **** indicate *** *** *********** very **** ** ********* an *** **********?

********* *** *** **** claiming ** ******* **** and, ** **********, ******* like **** *** **** learning. *** **** ****** answer ** **** **** is **** ******, *** is *********, ***** ****** is *** ******* ******** in **** ***** *** we **** ***** ******** reports ** ****** ******** unbranded ***** ** *** past ** ***** ********* like *********.

**** *** *** **** likely ******, ** *** mean ****** ********* ***?

***** * *** ** the ******* (********* ***) Chinese *** ********* *** claiming ** **** ******* SoC, ***** *** *** don't **** ** ** going *** **** ********* this. ** * ******* I ** ********* ** they **** **** ****** in-house ******** ** ***** they **** ** **** stock **** **** ***'* need ** **** ****** changes.

***** **** **** ** much ***** **** **** don't **** ** **** rushed *******.

***:****** ********* ******** ******* Surveillance *************. ** ******** *******, the ********* ** *** China **** ** **** Dahua *** ********* *** well ******* ** ********* chips.

******** ** ********* ** the ****** ****** ***** is * *** ** issues, ***** ***** ***** to **** ****** ** certain ***** (**** ****** non-HiSilicon) **** *** ********* lots ** ****** ** certain **** ******** ** HiSilicon.

Read this IPVM report for free.

This article is part of IPVM's 6,660 reports, 897 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