opticalflow

WOW, well-played, sir, very well played indeed. I may have to steal your snark.

> Actually looking forward to MS's UX folk cleaning up LinkedIn.

Wait, what? Are you saying you want Microsoft to "Metro-ify" LinkedIn? Forget the OS, even, for a minute... Just look at what they did to Skype and Office.

Are you saying you LIKE their current UX ideology?

It certainly worked for Edison, didn't it?

Just curious -- why would you think not?

What's also impressive to me is that Google's solution seems robust, meaning it does a good job in the face of outliers like lens flare, exposure changes, and the like. I see a lot of research with canned results where they optimize for the 6 or so clips they test with, but it's an order of magnitude harder to make it robust enough to release it as GA to the general public.

I'm an advocate of publishing not just the paper, but the compilable source and executable binary code and datasets used for just that reason. Science is not science if others can't reproduce the results and tweak and permute the inputs, reliably and repeatedly, IMHO.

I would posit that a deep learning network that learns to optimize parameters for a complex algorithm outside the convolutional network itself may have immense utility outside the classification problem. Call it a marriage of classic computer vision with deep CNN, or a hybrid approach. I don't think it's a binary decision. A deep CNN can find the optimal parameters (once trained) for a classic CV problem for a given image or video or other dataset, like superresolution, patch-based inpainting, or motion tracking. The training is the most computationally intensive part. As someone with way too many kids, I can testify...

There's some overlap, but what impresses me is that they're doing camera tracking (hard) in near-real-time and automatically (even harder) on an iPhone. Camera tracking is used in feature film visual FX frequently, and it's usually a very manual and specialist intensive process with very esoteric (and slow) tools...

That's assuming Zavodnik sent the requests via certified, registered mail. Zavodnik could have easily scribbled the address in unreadable gibberish and slapped a simple 1st class stamp on, in which case they would never have actually been delivered. Costello claims he never received anything.

This proposal sounds great, but seems to wave its hands furiously about right-of-way and eminent domain. I used to live in Astoria on Ditmars Blvd, and the proposed line would most certainly be going on/over thousands of private residences. I presume their placement is to try and connect this to the end stop of the N subway line, but It's not like Astoria is a junkyard or railyard, it's a dense residential neighborhood with single family and multi-family housing.

What could possibly go wrong with installing DOD's Root CA? Uhmmmmmmm

Clearly they'll need a 1.1 just in time for Oct. 31.

This looks to me like something on the order of convertible warrants based initially on bitcoin/Ethereum, which convert to DAO tokens. DAO tokens appear to be some sort of collective ownership of "crypto assets" whatever THAT means. The website reminds me of a horse with peanut butter in his mouth -- his lips are moving but not much is being said...

Very thorough article, with lots of good research and information.

Well, I think you're on the right track. Right behind me are 30 supermicro pizzboxes (some have GPUs) that are racked but powered down and looking very lonely (but with 6 dark fibers also feeling lonely at the moment). Perhaps I'd consider an "encoding.com for VR". Honestly, the main part I'm missing is real-time stitching, but maybe Facebook's open source release for their Surround 360 software later this summer will solve that!

All very good considerations. The "live" VR 360 capture market is really just getting started. I like the idea of "post-processing" as a service, especially. It might be down the road a few months though as the market for it develops.

Author of TFA here -- it was less about getting noticed than my not wanting to let the technology and possibilities languish in our back pocket. It was less relevant a couple years ago because the 3D market kind of crashed, but more relevant today because of VR.