Axis CTO: Deep Learning 'Exciting Space' But Urges Caution

By John Honovich, Published Dec 27, 2016, 08:54am EST

Deep learning interest is rising in video surveillance.

2016 saw:

Now, Axis CTO joins those touting deep learning but also advises caution about its adoption and timetable.

Deep ********

**** ******** ******* **** the *********** ******** ** performing ***** ********* ** video ************. *************, ***** analytics **** ********* ******* objects *** **** ********** (like ****, *****, *********, etc.) ** ***** / assess ******* ** ****** was * *****, *******, animal, ***. ** ********, deep ******** **** ****** networks **** ****** ******** data ** ******* ******, vehicles, *******, ***. ******** / ******** ******.

*** **** ********* ******* of **** ******** ** that ** ************ ** **** **** accurate **** *********** ***** analytics. *** *******, **** is * ********* ***** from ****************** ** ** ****** event:

IPVM Image

*** **** *********** ******* with **** ******** ** that ** ** **** more *************** ********* ** train *** ******* **** approach, ***** ********* ******** costs *** ********* **********.

*** ***** ***** ** from * ***** ************ series ** **** ******** / ****** ********:

Axis *** ** **** ********

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

*** ************ ******** *** an ******* ** *********overpromising **** ***** ********* and I’m especially conscious of that that when it comes to deep learning. While we have been developing *** ********** ** **** *** ***** **** ****, I’m not ready to share too many details. I think deep learning has ** ****** ******* ****** ** ** ***** *** ******. So, while some of our competitors *** ** ******** to offer deep learning, Axis is committed to making **** *** ************** is robust, reliable and addresses real challenges. [emphasis added IPVM]

** *****, ***** ********* has ******** **** **** than * ****** ** overpromising. ******,**** *** ***** ********* offering ** **** ****** quite ******, ******** **** ******* ** OEM ****** ** ****. ** ******, **** is ****** *** *****, as *** ******** ** filled **** ****** ***** analytic ********, **** **** large *** ***** *************.

Outlook **** ********

** ** ***** **** Axis' *** ** *** caution *** **** ********. While **** ******** ** rapidly ******** *** ***** new ****** ** ***** surveillance, ** ***** ** actual*************************, *** ******* *** rare. ********, *** ********** products **** **** ******** are **** ********* ** even ********.

********* ********* ******* **** learning *** **** ****** as ** ******* ** build * ********** ****** being *** *** ****, Chinese **********-********** ******** ** SMB ******** ***************. *******, as*** ****** *** ******** Hikvision's ******** ******* ******, ** ** *******, at ****, *** **** translates **** * ********** ready ********. ** ******, Hikvision's ***** ** ******* production **** ******** ***** analytics ****** ** ******* closely.

Conservatism ** ****

** ********, **** *********** culture ** ************, ********** their ************** ******* ***** not ********* ******** ** poor ******* ********. *** example ** *** **** - **** ***, **** was **** ********* * full *.*** ** ****. In **** ****, **** rapidly ****** ** *** surpassed **** / *** of ***** ***********.

*** ********* *** ** different. ***** *** ******* video ************ ************* **** a ************* **** ***** record ** ******** *** video ********* ***********, **** have * **** ********** approach ** ********* ***** and ** *** ****. If ***** **** ******** video ********* *** **** enough, **** ***** ** severely ******** ** ****.

** ** *** ***** for ** ** **** where **** ************ ** at **** **** ********, given *** **** ** actual ******** ********, *** as **** *********, ** is ********** *** ** the ****************** ***** ** *****.

Comments (4)

There is mammoth potential in deep learning and I hope Axis and others are able to deliver a more effective product than Hikvision.  The search for video events has to be as good as Google Image Search and make use of the same methods as Google to search the events.  Manually typing "Vehicle=true, Quanty=2, Color=red" does not pass as a user friendly query. 

It is very interesting article and makes me review my surroundings.

Deep learning of HikVision was very big impact to me, and I/We started to investigate for deep learning actually. And I also understand Axis's concern of unmature of technology,  but it is quite right that  this time is to prepare for future.

I think the view point for this issue could be changed to customer from manufacture. If customer may accept new technology even though it is immuture, it is valuable and could be a start point of development and investment.

“AXIS: I am a dire wolf, prey-stalking, lethal prowler.

HIKVISION: I am a hunter, horse-mounted, wolf-stabbing.

AXIS: I am a horsefly, horse-stinging, hunter-throwing.

HIKVISION: I am a spider, fly-consuming, eight legged.

AXIS: I am a snake, spider-devouring, posion-toothed.

HIKVISION: I am an ox, snake-crushing, heavy-footed.

AXIS: I am an anthrax, butcher bacterium, warm-life destroying.

HIKVISION: I am a world, space-floating, life-nurturing.

AXIS: I am a nova, all-exploding... planet-cremating.

HIKVISION: I am the Universe -- all things encompassing, all life embracing.

AXIS: I am Anti-Life, the Beast of Judgment. I am the dark at the end of everything. The end of universes, gods, worlds... of everything. Sss. And what will you be then, Dreamlord of Deep Learning?

HIKVISION: I am hope.”

-A pun with a twist deviated from Neil Gaiman's, The Sandman, Vol. 1: Preludes and Nocturnes

Its hard to argue with that. Very hard.

Read this IPVM report for free.

This article is part of IPVM's 6,728 reports, 907 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