Ex-Googler Startup Camio Cloud Video Tested

By Ethan Ace, Published Dec 16, 2015, 12:00am EST

Two ex-Googlers, one who was a Director of Product Management and another who is a Stanford PhD, have built a cloud video solution that aims to deliver Google style searching to cloud video / surveillance.

Camio has introduced advanced analytics (people, vehicles, color) and plain language search, essentially allowing users to "google" their Camio cameras for the video they want.

This search is shown in this animation:

We tested Camio with IP cameras, including Axis and Nest, to see how well these features performed.

Exec *******

*****'* *********, ***** ******** search, *** ***** ****** features *** *****, ***** in ***, ** ***, competitive ***** *******. ***** language ****** *** ****** accurate *** **** ******* ***** easier ******** ** ******** or ***** ******** ******** by ***** *********, **** professional / ********** *****.

*******, *** **** ** live ***** *** **** cameras ***** ***** ******** for ****. ************, *** setup ******* ******** *** most ** ******* (***** from****/********** *****) *** ** difficult *** ************* *****.

***** *** * **** powerful *********** ********* ** ***** search *** ** ********** solution *** **** '*******' users *** **** **** setup *** *********** **** video **********.

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

** *** ***** ****** / ******** ****, ******* may **** **** ** consider:

Key ********* ********

  • ***** ********** ** **** IP ******* *** ***** clips/snapshots, ********* ***** ** manually ****** ******* ** the ****** ** ****** on ******.
  • ****/******* ***** ** *******, only ********* ***** ** log **** ***** **** account *** ****** *******.
  • **** ***** ** *********** on **** ** *******. iOS/Android ********* ******* ******* live *******.
  • ***** ******** ****** ** simple ** ***, **** users ****** ******** **** they *** ******* ***, e.g., "people ******* **** ** office ** *******." *************, searches *** ** ***** via * ***** *** click ****.
  • ********* **** ****** ********, with *** ****************** ** objects, *** *** ****** classifications.
  • ***** ****** ***** ******** a ****** *** ** scan ******* ***** ***** may ********* ** ******.

*******

*** ** *******, ***** starts ** $*.** *** camera *** ***** (****** "Plus"), ***** ******** ** day *******, ******, ******, and *** ***** ******. This ******* ** ******* to **** *-** *** cloud ******* ***** **** competitors **** ** ****. *** Pro ****, $**.** *** camera *** ***** ******** 90 **** ** ******* as **** ** *****'* advanced *********.

***** **** *** ***** any **** ***** *** live ********* ** *** and ******* ****** *******, with ** ******* *** search, ** ********.

*****

***** ***** ** ********* compared ** **** ***** services. ***** ********* **** IP ******* ** ******** rules ** **** ***** clips ** ********* ** motion. ***** ******** ****(*), FTP, *** ***** ************ of ***** ******.**** ***** ****** ******** *** ******* models, *** ******* *** not ******** *** **** manufacturers.

***** *** ******* ** "10MB*** ****[*] ********* ***", ***** ***** ********* as "*********, ** ******* of *****+***** ** ** frames/second ** ******* **********." In *** *****, ** were **** ** **** higher *********** ******* *****, though *********, **** ** dependent ** ****** ***********, framerate, *** ***** ***** settings.

***** *** ****/******* *** much *******. ***** ***** their **** *********** *** select ***** ******(*) **** want ** *** **** Camio. ** ******* ***** is ********.

** ****** ***** ** this *****:

Plain ******** ******

***** ****** ***** ** search ***** ***** ********, such ** "****** ******* **** near ****" ** "**** approaching **** ** *******." This ***** ** ****** for ************* ***** ** find *****, ******* ** using **** ******* ******** or ***** **** ********.

**** **** **** ***** in **** ****** ****'* required; ***** ****** ******* them. *** *******, ** "****** ******* ******** ****", **** ****** (*** analytic ****), **** (*****), and **** (********** ****) are ****. 

** ****** ****** ** this *****:

Daily *******

*** *** **** *******, Camio ***** * ***** update ***** ********** ***** of ********, ********** ** what *** **** *** marked ****** **/**** ** the ****, ***** ** interest, ***. *** ***** contains ** ******** .*** of *** *****, ***** users *** ***** ** launch *** ***** ****** interface.

**** ***** ******** * simple *** ** **** through ****** ***** *** have ********* **** ******. However, ** **** ******, it *** ** **** useful, ****** ******* **** activity **** **** *** same ** *** ******, making ** ****** ** determine **** **/** *** important.

*** *******, *** ***** contained ***** *****. *****, because * ******* ********** the **** ** *** lower ****:

 

*** ** **** ****, because * ****** ****** through * ***** ***** was ********* ********* ***** or ******** **** *****/********.

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

**** ***** ** *********** for ** ******* ** Camio. *** "*******" **** shows **** * ************ updated *********, ** ******* of *** **** ***** from **** ******. **** video **** ******* *** iOS ******* ** *********.

***** **** **** ***** ** intended *** * ****** release.

IFTTT ***********

***** **** ******** ******* *******, *** ***** ** many cloud ********. **** ****** users ******* ***** *********** of ***** ** ***** systems. *** *******, ***** may **** ** ** outside ***** (*** **** or ***********) **** * person ** ******** ** a ******** ***** ****** zone, ** * **** may **** **/*** ********* and ****** ************* ***** on ***** ********.

Comments (9)

Undisclosed #1

IPVMU Certified | 12/17/15 04:40pm

That was quick!

Natural Language Scene Processing: The Future Of Analytics?

Apparently they were still at Google while developing this so maybe Google still has some financial interest?

Agree
Disagree
Informative
Unhelpful
Funny

Undisclosed Manufacturer #2

12/17/15 06:07pm

Could this level of AI be beyond what VMS companies develop? They just may not be able to match Google in this arena.

The question I ask is, is there some way VMS products can integrate with these types of services?

If not, as the AI behind these types of services becomes more and more sophisticated, what will VMS companies do?

Agree
Disagree
Informative
Unhelpful
Funny
Avatar

Carter Maslan

Camio | In reply to Undisclosed Manufacturer #2 | 12/18/15 06:17pm
I'm Carter, co-founder of Camio. Yes, we very much want to partner with VMS/NVR/Camera makers and will provide firmware source code and white-labeled apps and services to make real-time video streams smart and useful with a blend of on-device and cloud ML/CV. Please contact me or partners@camio.com.
Agree
Disagree
Informative: 1
Unhelpful
Funny

Undisclosed #1

IPVMU Certified | In reply to Carter Maslan | 12/18/15 10:19pm

Hey Carter!

I'm thinking maybe I connected two dots that are unconnected up above when I gave that link to the Google/Stanford research on natural language image searching.

Is there any connection or is it just coincidence?

Agree
Disagree
Informative
Unhelpful
Funny
Avatar

Carter Maslan

Camio | In reply to Undisclosed #1 | 12/19/15 01:24am

Camio isn't yet using that particular research for natural language descriptions of events, but NLP labeling is on our roadmap. Our current focus is on the precision of labeling and alerts at the lowest cost possible.

Agree
Disagree
Informative
Unhelpful
Funny

Undisclosed #1

IPVMU Certified | In reply to Carter Maslan | 12/19/15 02:35am

...but NLP labeling is on our roadmap.

What's the difference between NLP labeling and what you are doing now?

Agree
Disagree
Informative
Unhelpful
Funny
Avatar

Carter Maslan

Camio | In reply to Undisclosed #1 | 12/19/15 02:53am

We currently uses true NLP only for the date/time query parsing so that Camio understands a query like [people approaching front entrance last Wednesday at noon] just as well as [people approaching front entrance 12pm Wed the 9th] because it truly understands ways to express date/time.

Our current video event labeling only approximates NLP by its combination of object, color, direction, zone, camera etc... so that it *feels* like NLP without really knowing full grammar/concepts that describe the event naturally. Our real NLP labeling will know that your query [When did UPS deliver a large package?] means "a person wearing brown approaching with a large package, leaving it, and then departing back to the truck." So it's a level of Natural Language Processing that's resilient to sentence structure changes because there's a deeper understanding of concepts.

Agree
Disagree
Informative
Unhelpful
Funny
Avatar

Tony Luce

In reply to Carter Maslan | 12/21/15 10:09pm

Hey Carter - we'd love to talk to you about collaborating. Hit me up at tluce@networkoptix.com.

Agree
Disagree
Informative
Unhelpful
Funny

John Honovich

IPVM | In reply to Tony Luce | 12/21/15 10:41pm

Note: We have a direct message feature now and you can message members directly by clicking on the mailbox icon next to their name.

Agree: 2
Disagree
Informative: 1
Unhelpful: 1
Funny
Read this IPVM report for free.

This article is part of IPVM's 6,910 reports, 923 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