Intel acquires Itseez, leading developer of OpenCV | Hacker News Reader

Intel acquires Itseez, leading developer of OpenCV

newsroom.intel.com

127 pointscpcat26 comments

> Intel is transforming from a PC company to a company that powers the cloud and billions of smart, connected computing devices.

Sounds like they want to go the same route as HP and IBM. I.e. transform themselves from a company that does cutting-edge research and makes universally useful products into a corporate blob with no specific purpose and niche products of dubious quality.

I don't think so. I think they genuinely want to be a computer vision leader, and provide a vertical computation/vision/connectivity stack like no other for developers to pick up.

Thing is, I don't see how the stuff from the article ties into their existing business. When I hear "IoT" or "self-driving car" or "Computer Vision" I think about ARM, Google, NVidia and a bunch other brands, but not Intel.

It seems like they're trying to catch up to the leaders of several newly fashionable IT fields while simultaneously pivoting away from their core business. That's why it strongly reminds me of IBM and HP.

Also, for the heck of it, compare the types of news you see on NVidia and Intel websites:

http://nvidianews.nvidia.com/

https://newsroom.intel.com/

Side note: I'm surprised Intel isn't focusing all their effort on the emergence of VR. Proper VR requires a powerful PC, which is one of the areas of computing Intel still dominates (along with NVidia). If they bought Oculus...

To be fair, Intel does have a history in computer vision. They were the original developer of OpenCV, back in the 90s.

As I understand it, Mobileye is the player to beat in this space. They're doing hardware that is specific to the task at hand and, IIRC, they include a good chunk of the software as well. Buying back the talent behind OpenCV would allow them to easily compete in this space.

Well, for one those self driving cars have a few Intel Xeons onboard. Those cars need more processing power than your phone can provide.

Obvious bolt-on to Intel MKL too.

I agree, perhaps they are following Google's cloud vision-ML like product line

I wouldn't mind another company like IBM, it means we might have more advances like Watson.

When you say advances, do you mean Watson the Jeopardy-winning AI or 'Watson' the nebulous 'brand' that IBM now calls a random assortment of APIs of varying quality? I agree with the former, but the latter is far from being an advance.

> transform themselves from a company that does cutting-edge research and makes universally useful products into a corporate blob with no specific purpose and niche products of dubious quality.

This is very true and funny only when read as a sarcastic comment.

I think it's a smart move. Every major tech company is investing heavily into VR/AR to hedge their bets. Examples:

  Facebook with Oculus(acquired for $2B)
  Google with Glasses, Cardboard, MagicLeap($542 million in series B)
  Apple with Metaio(acquired for ??)
  Microsoft with Hololens
  Sony with PSVR
  HTC/Valve with Vive
  Amazon with Flow app
  Snapchat with their filters
  Uber with self driving cars

Basically every major company out there is doing something with AR or VR and they will all use OpenCV.

Also despite the fact that a lot of computer vision algorithms are GPU accelerated, the vast majority of them (at least in opencv) runs on CPU only. Intel can definitely do a lot of great stuff there.

Back in March, Intel bought a company by the name of Replay[0]. Replay used what they called "freeD" which uses 28 cameras around a sporting arena to capture a "bullet-time" style image. It takes some time to stitch this image together, but they can then use this image to fly around a player in the TV broadcast.

I could see this system using OpenCV to help decide what info is important in the image.

[0]http://www.usatoday.com/story/tech/news/2016/03/09/intel-buy...

edit: forgot to add the link

Google have a much bigger play now, they've built VR into Android N along with reference headset/controller design and apps like YouTube specifically optimised.

I didn't follow that piece of news closely but I thought it was just native support to make products like Oculus's GearVR more usable but you will still need something like gear vr to make it work?

You'll still need a headset and they haven't put any details out yet, but I think they're standardising the remote. The main thing though is they're building a full VR experience in Android itself and that seems like a much bigger play than Cardboard ever could have been.

Funny to see Intel re-take control of OpenCV after handing it off(?) to Willow Garage and then Itseez. Guess it turned out to be more important than they initially thought?

(?) Don't know the details of the transitions and why they occurred

When Gary Bradski worked at Intel, he was the leader of OpenCV. He was the only one there who was passionate about it and Intel officially supported the project basically as a favor to him. So when he left to found Willow Garage, they saw no problem with letting him continue to be the leader of the project.

Pretty sure they are just trying to bring as much knowledge as they can inside Intel. Plus it's good for publicity amongst the computer vision crowd.

I expect little to change.

Ok, any guesses on so how would this affect their open source projects? i.e. OpenCV's license is 3-clause BSD, and of course as Intel started the project is a copyright holder, so things as usual?

I wonder what impact this will have on their Accelerated CV library.

http://itseez.com/products/accelerated-cv/

It would be nice if this and whatever iOS equivalent they presumably have in development became free or cheaply available. It seems like this might not be in intels best interest though.

They also own a big stake in Tobii, leader in pupil tracking.

Interesting to see them positioning themselves for the future here.

The strategic case for it is that it lets them optimize the software for their hardware and vice versa - same reason they have C and Fortran compilers with a reputation for generally good number-crunching performance.

I wonder to what degree the rise of CUDA played into this. A lot of CV tasks are better performed on the gpu and today nvidia is the goto platform for that.

Yeah, they have a parallel strategy (for now) with their Xeon Phi hardware, but there's still a performance penalty vs Nvidia. A big problem is that all the software is written for CUDA and optimized for particular Nvidia cards - hence, software / hardware synergy.

Wouldn't be surprised to see them acquire more deep learning algo startups as well.

They missed the ball by letting the mobile market get picket up by ARM. the internet of things is coming and its going to be the internet that sees things!

Reply on news.ycombinator.com