Jesus Christ please no. Does anyone really think this is a good idea?

Addendum: of all the people replying to my post, saying how great it could be, I'd be willing to guess that none are in positions where they get harassed on the bus on an almost daily basis (eg being a woman).

If this happens, I suspect it will change human interaction significantly and make it more different to be casually anonymous. I'm not so worried about people I casually interact learning my identity; I'd be more concerned about cradle-to-grave government collection and permanent storage of these records, coupled with license plate scanners, etc.

As far as recognition goes, most modern algorithms don't use fiducials (what you call 'nodal points') to do recognition. So there's nothing in our code that has to do with interocular distance or mouth width, etc., so all of the features are "learned". However, fiducials are used to do a frontalization or re-alignment step before passing it through the feature extractor (neural network or otherwise). This type of frontalization step allowed Facebook's deepface algorithm to use locally learned filters for their CNN as opposed to shared weights.[1][2]

[1] http://www.cv-foundation.org/openaccess/content_cvpr_2014/pa...

[2] http://mmlab.ie.cuhk.edu.hk/pdf/YiSun_CVPR14.pdf

[3] http://yann.lecun.com/exdb/publis/pdf/chopra-05.pdf

One of the earlier successes of face recognition as described in Pentland and Moghaddam's paper was to use simple PCA on the images of faces normalized by eye position. This PCA would use all the pixels of a vector made from an image. Obviously there's been a lot of progress since then, but all pixels are getting used for information.

Principal Component Analysis looks interesting, but I'm struggling to understand how 2D pixel data would be transformed into the appropriate covariance matrix. Any insight on the missing link you could offer?

The technology has advanced far beyond that, but that was the start where automatic face recognition first became realistically possible.

For example, you're at a party/mixer, so you set your locality privacy settings to show your name to those in proximity, or to those in proximity that you have a direct or once removed connection to through Facebook or G+.

Another example, you are at a conference, so you set the locality privacy filter to show your name, location, and workplace to anyone in your work address book that happens to be close by.

Finally, imagine you are on the bus, going to your good friend's house, who you've set to always see you when within a mile or so. You see a marker for her/him show up in the distance, rapidly approach, and pass you going in the other direction. You realize they misunderstood your prior conversation about meeting, and are on their way to your house, or at least, were, until they realized you were on the bus they just passed and turned around to pick you up at the next stop, which you've quickly agreed to get off at.

I think there are cases where capabilities like this could be extremely useful and desired, it's just a matter of working out the issues.

This, and a thousand other uses, are what people mean when they say "It could" with respect to AR enhancing our everyday life.

When the government does facial recognition, they're not asking you to opt-in to being facially recognized... why do you think that is? When the police deploy license plate scanners, they don't ask you to opt-in and provide data, why do you think that is?

Literally the first app developed on a system like this would be something to "look up and stalk hot girls you see everyday".

Or maybe it's any one of a thousand other possible solutions, or capabilities. I'm really just trying to illuminate that when we are talking about something as far reaching as enhancing one of our senses with network information, and the unknown unknowns are enormous, I'm really just espousing a bit more caution before it's written off as useless, or a toy, or not worth the hype we've seen. I don't think we've seen a fraction of what's possible yet in this space.

No, there aren't. It is not possible for a system which is opt-in to get off the ground.

Do you want to purchase Google Glasses, which are opt-in (and which therefore identify 0.000001% of the people you see every day), or China Blackmarket Glasses, which are opt-out, and which not only cost 1/10th as much, they also identify 99.735% of the people you see every day, along with their birthdate, home address, cell phone, relationship status, sexual orientation and SSN? Which product will be successful?

No one is calling such things "useless". They have plenty of "uses". I'm calling them disastrous, and abusive, and privacy-invasive.

Nor is anyone saying they won't happen. They will, undoubtedly. And it will be ugly.

Opt-in was your assumption. Even so, that statement is trivially proven false. Facebook is opt-in by nature of needing an account. There's plenty of adoption.

> No one is calling such things "useless". They have plenty of "uses". I'm calling them disastrous, and abusive, and privacy-invasive.

Well, a statement close to calling AR useless (for these situations) farther up-thread is what started this.

I think the world will just adapt as needed to the new reality. Too easy to identify someone when in public? The public will adapt norms that make that harder. Right now we have assumed implicit anonymity, which isn't really anonymity at all in many circumstances. I'm not sure moving to a situation where people think about anonymity and how much of it they want when going about their lives, and then taking steps to try to achieve that level is entirely all bad. To be sure, it's got it's downsides, but I don't think ignoring the upsides and assuming that society will stay static while all this changes is unrealistic.

Even better, facial recognition plus a decent method of determining height, weight and ethinicity, and you have an x-ray specs app. The first person to make that work is probably going to be a millionaire.

Perhaps billionaire, unless the current startup bubble bursts first.

Let's see:

   LinkedIn -- get an idea of income via job category
   Tinder -- hmmm, someone "looking for love"
   Uber "walk of shame" -- someone not just looking

But in small communities, everyone already has this feature (by virtue of their being few enough people that an unaided human brain can track them all). And in small communities, you don't get a lot of jerks going around randomly harassing women on a daily basis. Why not? Well, the identity of the jerk who does the harassing is also known, and social (and ultimately physical) sanctions will be applied.

Now you're saying that in a large community where people are anonymous, there are jerks going around harassing women on a daily basis. But that's in a large community without identity tracking, where the jerks don't fear reprisal.

So what you're saying is actually an argument in favor of implementing AR identity tracking and letting human capabilities catch up with community size.

Oh, and I that my consent is required before somebody does face recognition on me.

Most of standalone digital cameras have an option to mute it, too.

It's not even remotely enforceable. For better or worse, this is the future.

But this specific example isn't the point. My point is that as information becomes more widely available and integrated in our lives, there will be plenty of ways to find out plenty of stuff about all but the most cautious hermits. Trying to plug every hole just isn't practical.

I don't see how a mandatory shutter sound is blocking "technological progress".

It's a bus/train/supermarket/DMV line, not a mixer. If it were a mixer I'd also be more interested in talking to the person than reading their profile and recent tweets.

I get how such technology would be impressive from a "oooooh cyberspace!" perspective, but that only reinforce's the point made earlier: Google Glass was very impressive tech that had a lot of technically interesting applications that were nonetheless mostly pointless.

Both my sisters have told me about borderline harassment and/or inappropriate attention they received while going about their daily lives.

You do realize most people consider that an anti-feature, and is exactly why a lot of people are worried about it, right?

I'm pretty sure this is a good reason NOT to give Glass to the masses.

Google, and hence the NSA and the US of America(thanks to Patriot Act) having the information of who is at every moment at a given bus, restaurant, party in any part of the word.... is kind of creepy to me.

With mobile phones they can do it today, but with lots of effort, it draws the battery and I could give my phone to someone else, or leave it on a Faraday box if I want.

With cameras pointing to people you have lots of other metadata, like who is kissing or smiling or hugging whom, who is serious with whom. Who is a leader, who is a follower.

In essence, it is a spy. It is not amazing, people from secret services regularly kill people from other countries if they consider it worth "national security", like a news reporter that wanted to know too much and know hold information that puts "the country at risk".

Talking to people like Arturo Perez Reberte that has been war reporters for decades they tell you very clearly that secret services from the big boys(Russia, US, Israel) have killed some of their friends.

Think for a moment how people, specially women preferred to live in cities instead of villages because they were not controlled in everything they did.

Businesses do that all the time. No one ever needs the next big thing, which is why they work so hard to make you want it.

What Google wants is to extend the capabilities of G+ into the real world. That's what augmented reality is, with Google behind the wheel. The problem it's meant to solve is the problem of you not generating marketable data while offline. Well, now you need never be offline. Now everything you see and hear is a part of the social layer. Isn't that wonderful?

This may be a 'significant augmentation' but it's not sure it's one that I want.

Recognition technologies currently in development can identify a person by gait, posture, and by face while disguised, turned away from the viewer, or otherwise obscured, albeit with a low probability of success. For now.

One of the few countermeasures shown to be effective against the newer recognition systems is "dazzle"-style camouflage. The other is to keep yourself out of any databases used for matching. Obviously, the former looks a bit outlandish for anyone out in public.

The paranoid and prepared will start scrubbing their social networks, wearing veils of some sort in public, and intentionally sewing rocks into their shoes now, so that when the technology matures later, they will be a "no match" to the databases already being built right now. Everyone else will screw their eyes shut and scream silently into their own skulls as their lives are pushed ever closer to resembling all the various sci-fi dystopias and utopias, blended and amalgamated into one terrifying and beautiful reality.