Meraki Object Detection Analytics Tested

By Rob Kilpatrick, Published Jan 20, 2021, 10:32am EST

Cisco Meraki's cloud analytics performed poorly in our 2019 testing. But now, they are claiming increased accuracy through machine learning.

IPVM Image

We bought and tested the MV72 and MV12WE examining:

  • Does it ignore common outdoor false alerts like animals, foliage, and shadows?
  • Does it ignore common indoor false alerts like shadows and lights?
  • Can it detect people running, crawling, or obstructed?
  • How far can it detect people?
  • Does it learn over time and reduce false classifications?
  • How simple is setup?

*******

***** ** * **** of *******, ******'* ****** detection ********* **** ************* improved **** *** ******** tests, ********** ********* ********** subjects *** ** **********/** scenes. ****** **** *** not ***** ***** ** common ******* ** ***** alerts, **** ** ******* foliage *** ******* ** lights ******* ** *** off.

*******, ******'* ********* ********** classified **** ***** *** small ******* ** *** scene ** ***** (**** common **** ****** *** camera **** **** *** legs *******) *** ***** rain ********* ***** ****** in *** *****. ******** parked ** *** ***** were **** ********* ******* and ***** ** *** analytics ***.

*******, ******'* ******* ******** ~15 *** *** ******** person *********, ****** **** many *** ********** ********* (typically ** *** ** lower), *** ***** **** Verkada's ****** *********, ***** required ~** ***.

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

** *** **** *******, ****** ****** ********* analytics ******* ****** ********* if **** ** ***** body *** **********, **** as ******* ****** * workbench, *****. *** ********* have ******** *** *** detect ****** ** *** lower **** ** ***** body ** *** *******.

IPVM Image

************, ****** **** ********** missed **** ******* ******* a ****/** *****. ****** are *** ********** ******** with ** ************ ***** used.

IPVM Image

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

** *** *****, ****** were ******** ******** *** classified ***** ******** ** running ******* ******* *** scene.

IPVM Image

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

****** ****** ******* ** false ****** **** *******, such ** ***** ***** changes.

IPVM Image

************, ******* ****** ******* the ***** **** ******* and *** ******* **.

IPVM Image

*******, ****** ***** ****** on *** *** ** other ***** **** ******* completely.

IPVM Image

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

****** ******* ******* ** false ****** **** ** swaying ******* **** *******.

IPVM Image

****** ******* ** *** ground ******** ** ******* were **** *******.

IPVM Image

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

****** ******* ***** ******* in *** ***** **** classified ** ******, **** as **** **** **** below.

IPVM Image

************, ***** ******* **** as *** *** ***** were **** ********** ** human.

IPVM Image

****** ******** ******** ** false ***************:

***, ** **** **** this **** ** ***** before. *************, ** ** not ******** ** ********* all ***** **********, ********** when ******* ** **** time ** *** ******, not ** * ****** or ** ** *** cloud. *** ******* ******** available ** * ****** constrains *** ******* *** DNN ***** *** **, and **** *** **** to ***** ********** ** some *********. **** **** of ***** ********* ** more ****** **** *** model *** **** **** to **** **** *.*., low ***** **********. ** try ** ***** ***** out ** *** ***** as **** *** ******** to **.

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

* ****** ****** ** false ****** *** **** analytics, ***** **** *** classified ** ***** ** Meraki.

IPVM Image

~15 *** ****** *** ****** *********

** *** *****, ~** PPF *** ****** *** person *********, ****** **** many ***** ********* (*.*.********,*****,******) ***** ********* ****** ~10 *** ** *****. However, ** *** ** lower **** *******, ***** required ~** *** *** detection.

IPVM Image

Stationary ******** ******* ** *********

****** ******** **** ********* counted ** * ******* that ******* *** ***** and ***** ** *** analytics *** *** *** camera, **** ****** ******** at ***** **** *** bounding *** *** ******* from *** ******* *** then ****** **** **.

IPVM Image

Default **** ********* ***** ** ************

****** *** *** ********* feedback ** *** ****** detection ******:

** *** *** **** model ** *** *** cameras, *** ** ****** vehicle ********* *** ******* cameras. ***** *** ********* neural ******** ********* *** these *** ** ******** from *** ** ** through ** *** ****. The ************ ***** *** a ********* ******* ************ and ** ******* ** a ********* **** ***. It *** ******* ****** than *** **** ***** in **** *********, *** worse ** ******. ***** is **** * ***** specifically *** *** **** fisheye ****** **** ****** for **** ********* ******* of **** ********* **** the ****** ** ** a ****'* *** *************.

** ***** ****** ******* that ******** *** ****** did *** *** *** of *** ***** ****** and ********* ** ******** animals ** ******.

IPVM Image

************, *** ************ ****** detection ********** *** ******* cone ** *** ***** as * ****** ******** times.

IPVM Image

*****

****** ****** ********* ********* are ******* ** ******* on *** ****** ******* and *** ** ****** in ******** *****.

IPVM Image

*** **** *** **** click ** *** ******/******* detections ** *** ********* tab ** **** ******* of *** **************.

*******

****** ****** ********* ********* are ******** **** * Meraki ******* ** ** additional ****.

Versions ****

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

  • ****: ***.*
  • ******: ***.*

Comments (0)

Read this IPVM report for free.

This article is part of IPVM's 6,805 reports, 913 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