Face Recognition Camera Tested (Netatmo)

By Ethan Ace, Published Jan 29, 2016, 12:00am EST

Can face recognition make home security "better"? That is the bold claim from startup Netatmo with their Welcome camera.

But with face recognition's historical challenges, can it really work in a sub-$200 consumer camera?

In this test, we take a look at the Welcome camera to find out, seeing how its recognition performs across over a week's worth of testing.

 

*******

*******'* **** ******* ******, its ****** *********** *** home/away ******** **** ********** in *** *****, ********** failing ** ********* ******** despite ********* ************ *************** for *****. 

******* ***** ********, ******* is *********** **** ******* consumer ******. ****** ***** and **** ******* *** simple *** ****** ****, other ******* ******* ***** functions ** ***** ******.

Key ******** 

****** *********** ***********:

  • **** ****** *********** ***********, with ***** ********** ****** or *** ********** ** known *******, ******* ********* *******'* placement ***************, **** ******** at ***** ***** (in **** *** *****. 
  • ****** *********** *********** ******* degraded ******* ****** *********, effectively *** ******* ** all.
  • **** **** ******** ****** *** camera ** ***** *****, well ***, *** ****** still, *********** *** ***** unreliable, **** ******** ****** faces ** ***** ******** not **********.

****/**** **** ***** ******, with ***** ****** *** as **** ***** * set ****** ** **** (4 ***** ** *******) instead ** ***** ** actual ******** ***********. ***** can ** ******** ****** as **** ************ ** as * *****, *** this ******** **** ************, limiting *** **********.

******** ** ******* ** a **** ** ***** clips ****, **** ** options *** ****** ********* or ******* *** ************ for * ****** ****, and ** ******* *** ********** recording.

************* *** ****** ** smart ***** *** ******* browser, ** ******* *** email. ************* *** ** turned **/*** *** ***** person.

*******/************

******* ******* *** $*** USD direct **** ******* [**** no ****** *********]. **** is ** **** **** other ******** ****** **** as ******* **************(***** **** ******** **** rec), **** ~$*** ******.

******* *** ** ******* subscription **** *** ******* or *********. ***** ** stored ******* ** *** SD ****.

Physical ********

** **** ***** ** review *** ******** ************ of *** ******* ******* camera:

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

******* ***** ***** *************** *** placement *** ****** ***********:

** ****** *** ****** in ** ****** ********, about *' **** ******** as **** ********, ***** 5' *** *** *****, well ****** ***** ***************.

******* ********* ***** **********, the ****** ****** ** recognize **** ********, ****** not ****** * **** at *** ** ********** known ***** ** *******. *** instance, ** * * hour ******, **** ******** approaching ~** *****, *** ****** ********** users **** *****.

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

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

******* ****** ************, **** as ***** ** *******, performance ** **** *****. The ****** ** ****** to ********** *** *** harsh ***** *******, ****** subjects ** *** **** FOV **************. **** ** especially ***********, ***** ***** positioning ***************, ***** ********* ****** the ****** ** * front ****.

**** **** ***** * subject ******** *** ******** via ** ******** ****, not ********** ** *** camera.

Identifying *****

***** ** ** ****** enrollment ******* *** *** Welcome ******. *******, ***** are ********** ** ******** on ******* ***** *** assigning * ****. ********* ** documentation, ******* ******** *** be *********, ********* ***********, by *********** ***** ******** times ****** **** ** shown ** *******. *******, despite *********** **** ***** >20 *****, ******** ***** were ********* ** *** lowest ********. 

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

Home *** **** ******** 

***** **** ***, ******* claims ** ******* ****/**** status ** ********** *****. When * **** ** seen, **** *** ****** as ****. *******, **** is ***** ****** ** a ************ **** ***** (1-12 *****), ***** ******** to **** *****, ******* of any **** ** ******** information.

***** *** *** *** problems **** ****:

  • *****'* ******* *** ***** elsewhere ** ****: ** * **** ** seen ** *** ******, but **** ***** ********* in/around *** **** ** goes ** ***, *** user ** ****** **** after **** ***** *******, despite ***** ***** *******.
  • ********** ** *** *** out:***** ******* ********** *** camera ** ****** **** an ********, ***** *** be ********** ** **** leave. **** ***** ******* may ** **** ** "home" ***** *** ***** expires, ******* ****** **** hours ***, ********* ** missed ******.

******* ** ***** ****** the **** *** **** features *** *********** * gimmick, ******* ** *********** useful.

Limited ******** *** ******

****** *** **** ****** are ****** ** *** camera's ** **** *** retrieved *** *** ***** interface (*******/******). ****** *** displayed ** * ****** list ** **** *** date **** ** ****** to ****** *** ******** events **** ** ** user, ******* *****, ****** motion. ***. ***** ** also ** ****** ** jump ** * ******** time/date, ******* ***** ** scroll ******* *** ****** to ****** ******** *********.

** **** * **** at ****** *** ******** in **** *****: 

Comments (4)

Interesting article

Agree
Disagree
Informative
Unhelpful
Funny

Another example of you get what you pay for.

Agree: 2
Disagree
Informative
Unhelpful
Funny

I agree and thought the (Welcome) indoor facial recognition camera was not up to expectations and did not perform as advertised. On the otherhand, the Netatmo Presence is for residential, short distance smart alerts (Cars, People, Animals, & Other Motion). It also has an adjustable "smart" LED floodlight. I have tested this camera down to -30F and up to 107F, in rain, high winds  and snow without an issue. After improving detection and providing user input to your scene this camera more accurate than the Nest IQ cam and has NO monthly fee!  I use the Netatmo Presence for monitoring my front and backdoor daily ($600 total MSRP for 2 "AI" cameras) and rely on it for accurate results 98%+. Traditional "Video Motion Detection" has disappointed me for years with 50+- false alerts a day.  In about 10-30 days of teaching the Netatmo camera it's errors; I have accurate alerts for my property after spending $4K+ on other solutions. Avigilon, Nest IQ, ExactVision, etc. It does have an issue with moving shadows and new objects introduced into the scene just like Avigilon, Nest IQ, ExactVision, etc. It will learn these false alerts and improve if you teach it. Also the Netatmo Presence does experience a 6-9 second delay in notifications and you cannot view current video until that notification is done recording. A huge bonus is the video analytic processing and all video storage (Micro SD card) is local to the camera. These cameras cannot be integrated into any VMS but are a great secondary camera with smart alerts to complement a typical 24/7 VMD 4K video surveillance system.

Agree
Disagree
Informative: 1
Unhelpful
Funny

#1, thanks for the detailed feedback and how you have gotten the Presence to work.

We did a consumer analytics shootout a few months back - Consumer IP Camera Analytics Shootout - Arlo, Google / Nest, Amazon / Ring, Hikvision / Ezviz

We don't have any immediate plans to test the Netatmo, mainly because, as you mention, it cannot be integrated into VMSes while it has very low recognition. If it did have 3rd party support (e.g., ONVIF, etc.), we'd definitely be interested in testing it as, like you have mentioned, there are a shortage of open quality analytic cameras.

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

This article is part of IPVM's 7,266 reports and 968 tests and is only available to subscribers. To get a one-time preview of our work, enter your work email to access the full article.

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