This year, the technology that people are talking about most is iPhone X’s new identification system that has replaced touch ID. This new identification feature is called Face ID – it identifies you and unlocks your device when you look at it. It sounds so simple, but how does this actually work?
The entire design and user experience of new iPhone revolve around Face ID. This is what let Apple ditch the Touch ID fingerprint sensor and their iconic home button. Most of the apps, including Apple Pay user flow has been redesigned around Face ID. Even the new Animoji entirely depends on Face ID sensors. So if Face ID doesn’t work, the whole iPhone X (including hardware and software) fails to deliver promise.
Today we’ll try to explain what goes behind the scene when you unlock your new iPhone. We promise we will keep it as simple as possible.
The Front Panel Combined With Advanced Technologies
Most of the equipment that face ID uses are located at the front part of the phone. Apple managed to squeeze a few different components here, besides a traditional front facing camera, microphone, ambient light and proximity sensor, you also have flood illuminator, and a dot projector, which is something Apple is calling a true depth camera system.
The last two components are essential for Face ID to work, so we are going to focus on them only. When you glance at your iPhone X, the flood illuminator lights up your face using infrared rays (invisible to the human eye). Once the system detects your face position, the dot projector shoots 30,000 infrared rays on your face in order to generate a unique 3D map of it.
Because this is all done with infrared technology, the system does not require any light to detect your face, it can perform in low-light situations without any issues.
The infrared camera captures the 3d pattern generated by the dots, which is further used to build a mathematical model of your face. This model is the matched against the one that was previously set on the phone. If they both are similar, your phone will be unlocked and you will be given access to the specific section you wanted to engage. All of this happens in real time.
The Chip Behind the Performance
The true backbone of the system is located inside the Apple’s A11 Bionic chip, which consists of 64-bit CPU with 6-core, of which 4 are energy efficient core, named Mistral, and 2 are high-performance core, named Monsoon.
Apple claim’s that it can perform up to 600 billion operations per second and it is used for machine learning methodologies, which are essential for face recognition process. Also, it utilizes the neural engine in order to adapt and learn about changes on your face, so that it can recognize your face when you are wearing glasses, change your hairstyle, grow mustache or beard, etc.
This not the first time someone has implemented face recognition technology, but Apple is the first to use this technology with 3D depth and neural network to get it done nicely.
Read: Inside A11 Bionic Chip | Apple’s Approach to Artificial Intelligence
Security
The Face ID data, including mathematical model of your face are stored locally in the A11 chip, not on the cloud. The data is encrypted and protected with a key available only to the Secure Enclave.
In order to improve experience, the data is regularly refined and updated as you use Face Id, including when you successfully authenticate. Face ID updates/refines the data when it detects a close match, but a passcode is used to unlock the phone.
Face ID matches against the depth information, which is absent in 2D digital photographs. It’s developed to protect against spoofing by pictures, masks and other methods using sophisticated anti-spoofing neural networks.
Face ID can recognize if your eyes are open and looking towards the screen, which makes it hard for someone to unlock your device without your knowledge (like when you’re sleeping).
If you are concerned about safety, the TrueDepth camera system is completely safe to use under normal conditions. It has been thoroughly tested and meet international safety standards. Since the output is low, It will not cause any harm to skin or eye.
Accuracy
The probability of random person looking at your iPhone X and unlocking it using Face ID is around 1 in 1 million (versus 1 in 50 thousand for Touch ID).
However, the statistical probability is different for twins and siblings that look very much alike, and also among children under 13, because their distinct facial features may not have yet completely developed.
Recently, Mashable posted a video on YouTube, in which they’ve showed a pair of twin fooling the Face ID. The video clearly shows that the iPhone X has a weakness for identical twins, it can’t really tell the difference.
Read: 15 Efficient Face Recognition Algorithms And Techniques
It’s not hard to guess, Apple is currently working on their Face ID system to make it even more powerful and accurate. In the meantime, identical twins will probably be locking their iPhone X with passcode.
