The iPhone X’s notch is basically a Kinect
theverge.com
theverge.com
Apple announced its purchase shortly thereafter and the Carmine got yanked from public sale ... thereafter Carmines were selling at a premium on Ebay, I suppose for competitors to reverse engineer.
Been waiting for this to crop up in an Apple product...
Incidentally it is a shame an equivalent device is not available to hack on...
[1] http://mashable.com/2017/09/14/inside-apple-a11-bionic-and-s...
Some sort of AI accelerator chip. Or clear phone / heads up display.
like the Neural Engine in the A11? :]
https://www.theverge.com/2017/9/13/16300464/apple-iphone-x-a...
The actual heavy lifting (model training) still happens and the """cloud""".
We were working with Authentec's 72dpi parts and complaining that the resolution was poor. In an engineering meeting, they told us that they had a high-res large-format part entering production. ("Large format" meaning "as big as your fingertip"; the existing high-res parts were maybe 5x5mm.)
Amazing! we said. They agreed to send us some documentation and samples.
A month later we had no samples. They denied ever telling us about such a part.
A year or so later they were acquired by Apple and Touch ID was the result.
But there are still a few depth cams out there other than the old Primesense ones or hooking up a Kinect to your PC.
WiFiSLAM in 2013
PrimeSense in 2013
LinX in 2015
Metaio in 2015
Faceshift in 2015
Emotient in 2016
Flyby Media in 2016
RealFace in 2017
and probably more that isn't obvious to me.
It was reported by Bloomberg, funnily enough in an article framed as Apple struggling in M&A [1], that Metaio took a lowball offer:
>Apple often refuses to work with investment bankers appointed by the seller, preferring to deal directly with company management, according to people who have been involved in such negotiations. Apple also dictates terms and tells targets to take it or leave it, betting that the promise of product development support later and the chance of appearing in future iPhones are alluring enough, the people said.
>That was the case when Apple acquired Metaio GmbH in 2015. Bankers appointed by the augmented-reality firm to negotiate weren’t allowed in the room, and while Metaio executives felt the offer was low, Apple’s vision for the technology convinced them to sell, according to a person familiar with the discussions.
>Apple’s current M&A strategy works well for acquiring startups developing new technology that can be added to existing Apple products. It bought 15 to 20 companies per year over the last four years. But buying larger companies presents a different challenge, particularly if there are rival bids. Bankers often diffuse tension between bidders and targets, but Apple’s approach can make that process difficult.
>“There’s a swagger -- you may call it arrogance -- about the culture there,” said Risley of Architect Partners. “They’re used to being able to muscle their way in and get attractive economics.”
Which seems completely logical on Metaio's part. It's obvious a lot of these startups working on fundamental technologies are just going to toil in obscurity, and selling to Apple is a chance to have your technology deployed and used in the biggest way possible.
[1] https://www.bloomberg.com/news/articles/2017-02-15/apple-str...
So yes Apple's main reason is for wireless charging but they also want to differentiate where they can and liquidmetal gives them "the hardest glass in a smartphone ever." (Ive quote).
Analysis and background including a very relevent patent application made a year ago and published in March for "using Liquid Metal (Metallic Glass) for the Backside of an iPhone":
http://www.patentlyapple.com/patently-apple/2017/09/apples-l...
In other words, the offer was realistic.
https://en.wikipedia.org/wiki/PrimeSense Kinect technology
https://techcrunch.com/2015/04/14/apple-buys-linx-a-camera-m...
https://en.wikipedia.org/wiki/Metaio
https://techcrunch.com/2015/11/24/apple-faceshift/
http://uk.businessinsider.com/how-emotient-ai-works-2016-1
https://techcrunch.com/2016/01/29/apple-furthers-its-vr-plan...
https://9to5mac.com/2017/02/19/apple-face-recognition-compan...
Contemplate if you will a hallway immediately past Customs in an airport. Equip it with multiple sensors of this type to provide complete coverage and redundant imaging/sensing. While you're at it, set up gait recognition as well. Then correlate the received profiles with the passport/identity data of the people just passed through customs.
Congratulations, you've just started building a database of everyone entering the country with biometric data that can be checked in the field with equipment costing less than $1000, and which can later be cross checked to find people traveling with false or multiple IDs ("this facial structure comes back as matching (person x) and (person y), and the gait is almost the same. Pick him up."), and it could all be done with technology that exists today.
Edit: autocorrect
https://www.google.com/search?q=unknown+comic&hl=en-US&prmd=...
Not that women in those societies can move freely, they must be accompanied by a man in most countries that do that to women.
[1] http://www.aljazeera.com/news/2017/09/brussels-airport-depor...
The current large expansion planned is capturing facial biometrics at exit, and using this to confirm departure (instead of relying on airline manfiests).
FB/Google jump on the train and enable new possibilities you don't want to miss, meanwhile Government gets nothing done, so in the end the three letter agencies will just arrive at the doorstep snagging all data anyway (formerly known as PRISM).
I mean why go to the trouble of installing expensive hardware identifying anyone if they all incriminate themselves for free on Snapchat?
Welcome to the new world.
The state may have a monopoly on violence, but private companies have more than enough power to ruin us even if they don't arrest us.
This thread has a lot of sensor speculation about what if scenarios. As a person working in FR and security, I can tell you we already have giant databases with millions of people in them. They are actively searched 24/7 with cameras located in public high traffic areas. Private spaces are filled with them. Retail is filled with them. You already see the cameras everywhere and do not think about it. We have systems that can search a 8 billion facial records on a single server in under 10 seconds, working with more realistic sized galleries, we have deployed and operating systems performing real time searching of everyone in a sports stadium or retail mall repeated 30 times a second for the duration of the public use of the space.
Like... I remember a few years back there was an article about using high framerate cameras to detect the heartbeat of people the camera viewed in hospitals. I wish surveillance could be used to watch people at risk for heart attacks, bring it to their notice if one is detected in the early stages, and direct them to their nearest hospital - in a way that involves a concert of technologies/devices.
Instead of CCTV cameras on every corner being used for criminals, use them to watch for health risks? Alert the person through their Apple Watch, send directions to the watch and phone, alert nearby first responders, notify family, etc... A route could even be established for an ambulance ahead-of-time, rather than as an ambulance rolls up to a signal.
I'm tired of worrying about terrorists. Maybe I want to be blind to the small chance a carbomb is possible. I know the most innovative stuff is often a result of defense spending/planning, but I hate thinking a driving force behind technology today is people afraid of other people.
</end-rant>
It seems to me that a significant part of the press and TV news are selling just that - fear. It doesn't take long to find a word like "threat", example taken at random, on the first page of the NYT ("Shinzo Abe: Solidarity Against the North Korean Threat").
If you subtract the cost of our insecurities and fears from the cost of living, life is actually really cheap.
Because amassing stuff gives psychological relief from such anxiety ("shopping therapy").
The same reason people on midlife crises go on shopping sprees...
>* Just imagine a world without any fear, just with different kind of markets and the same amount of money.*
Not all possible worlds have the same level of shopping activity.
It took heavy advertising and quite a lot of conditioning and pressure for early 20th century society to turn it into the shopping culture that it has today.
News do fear mongering for the same reasons people write click-bait article titles - it works.
It works because humans are not that far removed from our ancestors - the primal buttons are the easiest to push and they work on the largest number of people.
Anyone whose job it is to compete for people's attention, knows this very well. To pretend there's people far above who are 'bad', or 'good' is a coping mechanism to avoid seeings people and life for what it is :)
Doesn't have to be some actual singular person.
Businesses, media etc work as collective entities, alone and together, towards their goals, that is optimizing their environment for maximum profit -- similar to how our cells, which are living things in themselves, do.
If profit is the basic measure of success for an organization, it will reward those individuals behaviors that help it grow -- to the point that a perfectly normal company, like WV, from seemingly decent people, will create execs that lie about their pollution emissions, bribe politicians, etc., and -- among tons of similar examples.
Similarly, it would try to expunge those in the organization that do moral or consumer friendly stuff that costs the organization money.
>Anyone whose job it is to compete for people's attention, knows this very well. To pretend there's people far above who are 'bad', or 'good' is a coping mechanism to avoid seeings people and life for what it is
People that do "what works" because it brings money in, other considerations be damned, are bad, end of story.
Everything else is a coping mechanism to avoid seeings people and life for what it is (and for them to rationalize to themselves that they "just do their job").
The point I was aiming for, is that people like what's easy, what's natural.
You're saying 'they're bad'. That's actually the Christian axiom from what I understand, that people are inherently sinful and bad. I'd like to suggest that feeling that way is a dead end.
In other words - I actually agree with everything you're saying, minus the 'bad' label. I label it 'I'd do it differently' which's subtle but very important. Labelling what most people around you are good with, bad, is simply going to make you a misanthrone [0] and that'll prevent you from leading the life you want :)
The point regarding coping - labelling people bad and good is an extreme simplification used for children. Once you're an adult, it's much more helpful to understand that what another does, you do in other ways, maybe much more subtle, and let go of harsh judgements. Judgements separate people into us vs them, which's what leads to most of the behaviour you're labelling bad in the first place.
It's a very subtle point, that most people endlessly miss.
Never said that "people are inherently sinful and bad". I said that (and I quote): "People that do what works because it brings money in, other considerations be damned, are bad, end of story". Which is a very different thing.
>Labelling what most people around you are good with, bad, is simply going to make you a misanthrone
Most people aren't profit seekers "other considerations be damned". And if they are, then they are bad -- doesn't even matter if they are the majority. You don't get a free pass on crapping on others because "most do the same".
>The point regarding coping - labelling people bad and good is an extreme simplification used for children.
Labelling a sad state of things as "natural" is an extreme perversion or an extreme manipulation. Human civilization is not based in whats natural. Some natural things and behaviors are good, others are bad. Civilization is all about making that distinction.
>Judgements separate people into us vs them, which's what leads to most of the behaviors you're labelling bad in the first place.
You gave me a $2 philosophy (and condescending at that, with the link to "misanthropy" to enlighten me of that "obscure" term), whereas I gave specific examples of behavior.
People promoting fear from their media outlets because "it sells". People who lied about emissions in WV's case. Etc.
Yes, we should absolutely separate people into "good" and "bad" for such offenses. I don't care if otherwise they are complex characters and "good people", e.g. great with their kids or tender to their adopted a one-legged pet hamster, when their actions hurt society for their benefit (be it money, promotions, esteem from their colleagues, etc).
It does create a different sort of "machines replacing humans" scenario. Instead of AI destroying us or replacing human workers because they are more effecient, machines replace us because they are just nicer people. They never get tired, hangry, frustrated. Superhuman levels of patience and empathy. I myself am a teacher, but I think I might prefer my children taught by TeachingUnit7000 who is just as excited about the lesson for the last class of the day as for teh first.
A simple checklist suffices for this. No need for AI camera surgery assistants.
cf http://content.time.com/time/health/article/0,8599,1871759,0...
> The study focused on six checklist items, all involving basic safety issues, [...] whether all the sponges used in surgery were accounted for after the procedure.
I will say however that the Apple Watch is now being (or soon will be?) used to detect heart problems and notify the wearer. So this kind of thing is happening.
Similarly crime watch apps do exist and could use the Apple Watch for notifications and safe navigation.
But if we had a crime detecting surveillance system there's no way to only use it to help good people stay safe. It would also be abused in ways that cause innocent people to suffer.
I'd just like to see a society that doesn't throw people in to jail (ruining their life) for trivial things. I mean, such societies exist - I'd like the USA to be one of them.
Many people are concerned. But as a matter of fact. More people die every single day in car accidents. We as a society should not give up our freedoms (the right to be forgotten, the right to remain anonymous) for the illusion of safety. As the recent London, Barcelona, or Charlottesville attacks have shown, you only need one crazy person to carry it out for whatever reason. And even with a sophisticated tracking network you won't be able to stop this effectively.
These acts happen because intelligence made a single mistake in a day-to-day work, not because terrorists invented yet another way to do it. The key points are communication and identification. No one will crash into a crowd without talking about that with co-crazies. It's also naive to assume that they prepare actions only few times a year. They really hate you, your life style, your everything. If you turn your defenses off, car accidents will appear like an insignificant loss in few weeks.
Why I'm stressing this? Because intelligence methods are naturally contradicting your freedoms. Blissful ignorance should not drive a freedom train.
[citation needed]
Upd: for more specific details:
https://thebreakthrough.org/generation_archive/effectiveness...
http://www.tandfonline.com/doi/full/10.1080/10242694.2011.65...
Just few of dozens with proper citation and research, for those who need it.
This one is for UK known as very serious in police regard, afaik: http://www.bbc.com/news/uk-39183003
And finally, this one is world-wide. Scrolling through 2016 on mobile may be pretty enlightening or frightening, depending on your views: https://en.m.wikipedia.org/wiki/List_of_thwarted_Islamist_te...
...You said "98%". It's right there.
>Additionally, the CCTV has had only a 3-15% success rate in identifying suspects of crime, burglary, etc. Given the generous assumption of equating the cameras' ability to stop crime to stopping terrorism, there would still only be a 15% success rate at the most. Counter-terrorism methods must be 100% effective or people will be murdered. This seriously impacts the argument that CCTV surveillance (as it is being done currently) would be able to effectively stop terrorism.
Hope these non-made up numbers clear some details on topic, my negative friend.
I'm not sure there's even any evidence it was planned at all (even 30 seconds in advance) versus a panic response to being under attack by a surrounding mob.
I'm not trying to defend any particular action, just that it does seem very distinct from, for example, a bucket of TATP and nails left in the subway. I personally am waiting for after the trial to draw any conclusion about Charlottesville.
Oh, please. Take one minute to watch the (graphic, disturbing) videos from the scene and observe that, before the attack, the Charger was nowhere near the "mob" when he made the decision to gun the engine to plow through it. Fields started his attack run from well down the street, perpendicular to the marching counter-protestors.
Why? Unless you have some kind of power over the defendant, there's no reason to reserve judgement until after the trial. Are you unwilling to make a judgement about, say, Julian Assange until he's tried in a court of law? Did you reverse your opinion about Joe Arpaio when he was pardoned? I have a lot of faith in the justice system, but verdicts are wrong all the time, and should be one of many factors in how you view the world.
But it's ok to think OJ was guilty, it's ok to think Arpaio is guilty, it's ok to have an opinion about Julian Assange even though he hasn't been tried.
The problem is that if random acts of violence are conflated with systematic terrorism, we grossly inflate the numbers in such a way as to be constantly crying out for further surveillance and erosion of rights. And when the attacks keep happening, because they're spur-of-the-moment lunatics, we will proceed to clamp down harder.
So, no, we have to draw the difference between the two.
Everyone's favourite source. https://en.m.wikipedia.org/wiki/Definitions_of_terrorism
However far right violence has a track record of not being labelled as terrorism quite as easily as attacks by Muslims. https://www.google.co.nz/amp/s/static.theintercept.com/amp/t...
I feel that largely we don't innovate for the good of society anymore - only to gain greater control over unpredictable masses that can riot over a Tweet. (((But mostly to catch terrorists & pedophiles.)))
1: https://www.adl.org/education/references/hate-symbols/echo
Three parens surrounding words like "pedophile" is something that shows up a lot on Twitter, and is meant as a dogwhistle for antisemitism - just so you know and don't get caught off guard of someone assumes that's what you mean in the future.
The issue is the fact that a "binary test" has 4 outcomes.
The good ones: Test says "True X" and "X is true". Test says "False X" and "X is false".
The bad ones: Test says "True X" and "X is false". Test says "False X" and "X is true".
Some of these can be problematic depending upon what "X" is, what the response to "X" is, what the probability of "X" is in the population, and what the probability of the "false positive" of "X" is.
In medical, this can pose a real dilemma.
At any individual test let's call the diagnosis rate 1%. If the false positive rate is about 1%, we are diagnosing one person who has "X" along with diagnosing one person who doesn't have it.
Not too bad, right? Ummmm ...
If "X" was heart disease, sure. Most of the interventions for heart disease are things like fix your diet, get some exercise, quit smoking, etc.
If "X" is breast cancer, not so much. Interventions for cancer start with something like a biopsy which can have lots of complications (hematomas, secondary infections, etc.). In addition, a cancer may not spread, or something else may kill the patient before if you left it alone.
So, the damage done in the response determines how low the false positive rate has to be.
In addition, there is a psychological aspect to a diagnosis that a lot of people overlook. A false positive for "heart issue" on your wristband can upset certain people so badly that it causes an issue.
And I haven't even started talking about the false negative rate problems. What will you do when people start saying: "Nah. I don't need to go to the doctor, my Apple watch isn't flagging anything."
This is why this kind of stuff gets deployed for "terrorism" first because there isn't any consequence to the manufacturers whether they're right or wrong--so they don't care about the relative probabilities. The fact that some random schmuck gets detained and searched isn't the manufacturers problem.
Creating something like a wide-scale database about individuals based on data collected without their consent or under duress could be treated, internationally, like the deployment of nuclear weapons in a first strike.
Likewise, we could charge people who deploy pop-up ads with crimes on the magnitude of a mugging. Botnet operators could be hunted with the zeal we currently reserve for drug smugglers and media pirates.
The reason people say that one has to lose at the expense of the other is because of weird "freedom losses" such as privacy. Lack of privacy is not a loss of "freedom". Surveillance, again does not mean a loss of "freedom". That is, unless you're doing something wrong, in which case the system is behaving exactly as it should (identifying someone that is doing something wrong). Tech-people constantly talk about old laws having to "change" for the new technological landscape we live in. But we rarely talk about changing how crime and justice gets handled in the new landscape. And it is a discussion we need to have, because we're moving into territory where automated systems can very easily detect and identify the occurrence of crime. Magically moving into "surveillance reduces our freedom" because it catches all of the commiters of that crime, which never used to be the case until technology enabled it, is a disingenuous path to take. Just look at red-light or speed-trap cameras to see how it's already been playing out.
Sure, once we start going into territory such as mandatory curfews and what-not in order to combat crime then you can start saying that we're losing freedom. We're not even close to that, as much as that pains me because I think that a government that permits the existence of easy-to-stop violent crime in 2017 is a morally bankrupt one that is complicit to those crimes on some level.
This seems like really screwy logic. How about we just don't spy on people? Or, if you think it could be used to help people, maybe start from a point of "how can we help people" rather than working backwards from "how can we come up with legitimate uses for mass surveillance".
I assume you believe this can be done accurately, or you wouldn't be suggesting this scenario. In that case—what exactly do you see as the problem here?
https://www.theguardian.com/australia-news/2017/jan/22/facia...
Customs will still delay you for not particularly good reasons.
Why do I need privacy, exactly? Sure, privacy from being spied-on without reason by random low-level government employees, sure. But I don't care about privacy from automated systems, or systems that are used in criminal investigations after the fact, or during the fact. Also, I'm fine not having privacy from detectives and criminal investigators that are investigating crimes.
That distinction is an important one, though, and is usually lost in discussions of tech and privacy.
I am pretty sure that each country already runs the same face but two different IDs check. Ironically, the poorer countries probably have better tech since they started fresh, countries like USA keep patching their decade old systems.
There are enough companies selling face recognition tech to states but Apple isn't one of them. So the iPhone X shows how much is possible but in itself won't change surveillance.
> The protein of eye lens is very sensitive to IR radiation which is hazardous and may lead to cataract.
Unless some technology is capable of killing more than few percent of population, government is the last to worry about statistics.
backscatter machines are considered complete safe, from what I understand. they use radio waves (not ionizing radiation) the same as phones/microwaves, and the waves don't even penetrate skin (or else they wouldn't be very effective, would they)?
you have nothing to worry about, medically, from them. I guarantee you're being bombarded right this moment with a similar amount of energy in a similar portion of the spectrum from Wifi, phone calls, bluetooth, FM/AM radio, GPS radio, etc.
https://en.wikipedia.org/wiki/Backscatter_X-ray
whereas you're thinking of these:
https://en.wikipedia.org/wiki/Millimeter_wave_scanner.
They are compating technoogies; the former employ ionizing radiation, the latter do not. I don't really know which ones are employed on US airports right now, but backscatter X-rays were employed at some point, which generated a bit of media shitstorm few years back.
> Workers in hot environments, exposed to IR, developed lenticular opacities due to IR irradiance in the order of 80–400 mW/cm2 on a daily basis for 10–15 years
That is really high. The original Kinect has an output power (at the emitter) of around 60mW. The expanded beam is safe to look at beyond a centimeter or two (I think less, actually) due to energy conservation.
On top of that you're only being exposed for a second or two to grab the depth image, not 5 minutes.
That was enough to put it off for myself for long term use after trying it a a couple of weeks, until I can find something that explains the risks - or rather, lack thereof - very definitely. I've attempted to search but have come up short each time.
The S8 iris scanner was super quick, very impressed by it.
(1400 watts / m^2) * 0.2 [ground albedo] * 0.5 [proportion in infrared] = 14 mW / cm^2 of infrared radiation looking at the ground on a sunny day.
(14 mW / cm^2) * 1 square foot [approx size of a face] = 13 watts
I'm going to guess there isn't a 13 watt infrared laser in the iPhone, or we'll be seeing some great hacks where people take the lens out and start setting things on fire from hundreds of feet away.
[0] https://www.sparkfun.com/products/14032
[2] https://en.wikipedia.org/wiki/Joint_Probabilistic_Data_Assoc...
https://cleantechnica.com/2016/07/29/tesla-google-disagree-l...
It is true that you may try to catch emotion from the user face but on the other side you can catch the world...
It still may be only good to a few feet out in direct sunlight, I don't know.
There are ways around the inverse square law (such as projecting a single/few laser beams) but the hardware becomes slower and more complicated.
The original Kinect is basically a one-camera stereo setup. The pattern is known and they perform correlation with a stored image of the pattern at a calibrated distance.
Basically use a surreptitious infrared camera, copy the face of a person, then project it onto a screen and I would mostly be good to go?
Alternatively, if I just plastered infrared dots on someones face, face recognition would cease to work? So 2 Iphone X could be used to interfere with each other?
I'm guessing The 2 sets of dots would overlap and confuse.
Actually on second thought I'm not sure it'll work. Your infrared ink dots will interfere with the projected infrared dots.
I also assume the dots are tracked to identify, so static dots wouldn't be considered, although it could cause problems.
I could see how using 2 phones could cause problems, perhaps they need to strobe them to avoid clashes.
And then the app developers upload the depth data to their servers and use it to track users, and then the servers are hacked and the depth data is taken by the hackers and then the hackers sell the depth data and then someone can use that data to unlock stolen iPhones. Sounds great /s
I don't know if they could have, but for something that just sits somewhere in your living room... that's a totally different design mindset than something that needs to fit in your pocket and whose main real estate you want to spend on a screen.
The Apple "innovation" here is finding an application for it: basically Snapchat. (Because, and let's be really serious here, FaceID is not a good application.)
...and that alone just shows how off the mark this implementation is. When this comes to the cheap Chinese Android phones in six months, it's going to cause another sales boom there, as parents will actually be willing to spend a few hundred dollars for a phone upgrade for a teenager, verses the thousand bucks for one of these Apple phones...
Why not? More secure face recognition seems like a very good application that will be useful to millions.
The basic working principle is probably also basic enough to have lots of prior art I guess.
Question is, will there be issues with face unlock in bright sunlight.
/s
If they used a variety of frequencies alternating at some known sequence or something it should be possible to differentiate signal from noise, but I have no idea what I'm talking about.
my issue with facial recognition on the phone is, will it try to unlock if I am just reading notifications/texts? not something I want it to do.
Think of it this way, what might be possible or normal once sensors like this are so cheap they come embedded in all OTS camera modules for even the cheapest of devices?
Pretty crazy right?
Even though I'm an Android user, I think I'm more scared about Google catching up and incentivising the high res scanning of faces for some consumer application.
I thought Tango included hardware to process parallax and phone position so it could potentially work outside and in bright open areas.
- Mckinsey's website http://www.mckinsey.com/business-functions/strategy-and-corp...
Earspeaker behind a screen works already (a Chinese handset manufacturer did this) and I dont need the other sensors and that for Face-ID. I found a fingerprint sensor better anyway.
I'm pretty sure Apple R&D is continuing apace :) They ship updates every year, so at some point during the cycle they have to decide what they'll ship in the next model. Figuring out which set of features to ship now that also keeps them on a path forward to a not-completely-decided future is quite a challenge, I'm sure. They've been pretty successful so far.
How did you find TouchID to be better than FaceID? Did you get a pre-release version of iPhone X?