Yesterday, Tuesday September 12, 2017, a rather sad thing happened: after the last flyby of Titan by the Cassini probe on Monday, NASA announced that it was going to carry out a final correction of its orbit on Friday so that its ashes are dispersed in the Saturnian atmosphere. A grand end for this valiant space soldier, after 20 years of good and loyal service. Well, otherwise, Apple held its Keynote during which the iPhone X was presented, with its new unlocking by facial recognition called Face ID. And since we know more about smartphones than space probes, let's see together how this new biometric lock works.
Apple iPhone X
Note TechnologiesTips (12) read the test
- eBay 114,80
- Back Market 320,00
- Certificate 324,80
- Fnac.com Marketplace occasion 328,00
- reBuy 370,99
- Darty Marketplace 385,00
- Darty.com Marketplace occasion 386,00
- Devil's Sale 429,40
- Rakuten 450,20
- Amazon Warehouse 517,22
- Fnac.com 809,10
- The Redoubt 964,98
- Materiel.net 1 202,89
- LDLC 1 203,89
- Cdiscount 1
- eBay 472,69
- Cellys.com 499,00
No arms, no chocolate. No iPhone X, no Face ID. For now, this new authentication method will only be available on Apple's most high-end smartphone – whether or not your current iTerminal switches to iOS 11, or whether you opt for the iPhone 8 /8 More or not. The limitation here is not software, nor does it result from a history of computing power, given that the next three iPhones will all share the same A11 processor. The limitation, for once, is materially justified. It is that, to do facial recognition, you need a little more than a "simple" photographic module on the front. The complete recipe is given to us through the demonstration video above, commented by the sweet voice of Jon "Johnny" Ive, and it is especially the details given at 1 minute 14 seconds that interest us:
Lots of people in so little space: components that are not underlined in green do not participate in Face ID.
To operate Face ID, the front photographic module does not work alone: it collaborates with the infrared module and a point projector. Everyone has their role. The main module acts as a conductor, the infrared module takes over when the light is insufficient and the point projector provides volumetric detection. But let's go into detail.
The advantage of an infrared module is that it eliminates the quantity and quality of light that illuminates your face. Even in the dark, he continues to see you. If it had been necessary to go through the main photographic module, in poor light conditions, it would have been necessary to increase the sensitivity, which would have distorted the detection. However, Apple does not provide more details on how infrared works: does it use an infrared projector or does it go through thermal face detection, according to a process used for a few years? For now, mystery.
The other key element of the device is the dot projector. And what is a dot projector? From Apple's promotional video, it looks like this:
With Face ID, you can finally pick up your phone without getting it wet. No more Claude François syndrome.
The little Minority Report-style dots are just there for illustration. Rest assured, in life, all of this will be invisible.
Your face is mapped according to 30 control points, this map then being sent to the processor for analysis, which will take care of reconstructing a virtual model of your face. If Apple puts so much emphasis on the machine learning aspect of its A000 Bionic chip, it is because the process thus finds all its interest. Indeed, the virtual model of your face is not static, but dynamic. According to Apple, it is able to adapt to your morphological changes, whether you gain weight after having abused the Sunday raclette (the season may start earlier this year), whether you sweat or have a tired face from having done too much sport (it is necessary to eliminate it, this squeegee), that you change your haircut, possibly facial hair, etc. Well, inevitably, if you wear a balaclava or pull your collar up to your eyes, it may complicate the matter a little: Face ID does not yet see behind the material. That said, it will be interesting, especially for bikers, to see how the iPhone X behaves when its user wears a helmet. And while we're at it, what happens when you have a high fever (admittedly, you're not supposed to check your phone then)? Or when you're chilled, say, you're a white walker, just sneaked out, and want to Snap to share the news with your friends?
Yole Développement analysts had, at the end of August 2017, very well anticipated the mapping technology used in the iPhone X.
By going through a triple confirmation in visible light, infrared and volumetric light, the vigilance of Apple's Face ID risks being a little more complicated to fool than that of the purely visual Samsung Galaxy (S8, S8+) and Note (7 and 8). A simple photo will not suffice. How about a 3D print of your face? Again, it shouldn't work, because it will miss the thermal confirmation. Afterwards, someone can really, really want to hack into your phone and go to the trouble of perfectly falsifying your head, in three dimensions and in temperature, but the chances of that happening remain quite low. Which doesn't mean Face ID is tamper-proof. But right now we don't see how — and even if we did, we wouldn't explain it. However, another question arises: Apple already knew everything about you, already stored up to five of your fingerprints: Cupertino will now have a dynamic mapping of your head. Here is information whose monetization could pay off big... Let's be reassured, however, all this information is currently stored locally on the terminals, without Apple being able to access it.
Face ID, therefore, should allow Apple to distinguish itself, at least for a while, from its Android competitors. The apple brand thus justifies its numerous recent acquisitions of ultra-specialized start-ups in imaging: the Israeli LinX in April 2015 (multiple camera specialist), the German Metaio in May 2015 (specialist in augmented reality and thermal interfaces), the Americans Perceptio and Emotiont (specialists in facial recognition) or the Swiss Facsehift, the American Turi and the Indian Tuplejump (specialists in machine learning), to name but a few. Will the German-American pmd tech be able to resist the voracity of the apple? The future will tell. In any case, this start-up did not hesitate to present, one day before Apple, its own three-dimensional facial recognition solution for which it offers a development platform intended for developers of Android smartphones. Well...