3D-Printed Heads Fooled Android Facial Recognition
- 2018-12-12 10:20
Recently, face recognition has become one of the best ways to unlock your phone or tablet. And even though the fingerprint sensors on the display are becoming more real and reliable, companies like Apple rely heavily on face recognition. Face identification is usually considered the best, but how easy is it to cheat? What about the best competitors in the Android world?
Forbes decided to figure out how to make a 3D model of someone’s head and test several phones to see if their fake head was unlocked.
Phones that participated in the experiment: iPhone X, LG G7 ThinQ, Samsung Galaxy S9, Galaxy Note 8 and OnePlus 6. The head cost 300 pounds sterling, or about 380 dollars, and in the process of its creation at least 50 cameras were required.
The head was printed in Backface in Birmingham, UK. Cameras come together all together to take one shot that makes up the full three-dimensional image.
So how did things go down? It’s pretty predictable, knowing how easy some phones can be fooled by simply taking a picture of someone’s face, but it turns out that all Android phones didn’t cope with the test and unblock themselves, facing a fake head. Anyway, Apple's iPhone X stood by itself as the only phone that refused to reveal its secrets.
However, if you are an Android user, look away from the screen. Four of the hottest phones with Google and Apple operating systems were tested to see how easy it is to hack them. This was done using a 3D-printed head. All Androids opened with a fake head model, the Apple phone, however, was impenetrable.
It's not entirely clear why the iPhone X refused to unblock itself, although our suspicion would be a demand for attention before Face ID unlocks the device. Whatever the reason, we are not sure that any of this is really important, at least for the vast majority of people.
We do not think that many of us need to worry about someone creating a 3D model of our heads just to see our text messages.
Leave a comment
Your comment is awaiting moderation. We save your draft here