We have all heard of the newest smart phones getting rid of Touch ID and implementing a new form of security called Face ID. Apple has it, Samsung has it, and other tech companies are not too far behind in the makings. Face ID is a super fast way to unlock your phone without having to worry about taking your gloves off, or freeing your hands from holding something. It also allows the screen of the phone to increase is size, without needing a home button. However, there are some disadvantages to Face ID, the major one being security. Every fingerprint in the world is unique, and Touch ID is much harder to fool than Face ID. With Face ID, it has been shown that identical twins can unlock each other’s phones, and even people who look very similar as well. With things such as Apple Pay, someone who looks very similar can make purchases on one’s bank account just though Face ID. Additionally, especially with the coronavirus, people are wearing masks, and Face ID is unable to recognize them, therefore forcing them to manually type in their password. Although the eyes can be seen, that is not enough to activate Face ID. Then how exactly is Face ID activated? Let’s take a deeper look. Face ID basically works with Artificial Intelligence (AI). When you first set up your phone, you have to enroll your face into the system. The phone takes a scan of your face and stores it, just like how you input your fingerprint into the system. With Apple’s True Depth camera system, your smartphone uses sensors and scans all of your facial features. These technologies work together to create a detailed map of your face that can be compared to. When setting up Face ID, the user has to move their head around a lot, and this is basically the system trying to catch every detail, to create a 3D map. Your phone uses infrared (IR) light to illuminate your face while creating the map. This is so that even when you are in the dark, in the light, wearing glasses, etc. the system can still recognize you. How exactly does this infrared light work? Essentially infrared light is a “non-visible illuminator” which means that it can make up for the effect of varying amounts of light or different conditions, and it will still be able to detect you. The interesting part is that infrared light is invisible to the human eye, so all of these infrared waves are bouncing off your face and you don’t even realize. After the ambient light sensor determines how much illumination is needed, then a part of the phone called the dot projector creates over 30,000 dots of infrared light. This maps all of your facial features, and the infrared camera captures images of the dot pattern. Next is the actual recognition part of Face ID. The human face is a biometric characteristic. Biometric means that it can be used to identify a person. The map of your face that is already stored in your phone is called an enrollment image. Then when you present the phone with your face again to unlock it, this is called a verification image. In the processor, once the face is seen, it is compared to the enrollment image. It then calculates a number between 0 and 1. If the number is closer to 1, that means it is the same person. However, if the number is closer to 0, that means it is not the right person. Although infrared light is used to make sure the lighting does not affect the facial recognition, the enrollment image and the verification image likely won’t be identical. The number the processor generates determines how similar the two images are. When just unlocking the phone, the system looks for a score around 0.6-0.7. However, when something secure such as authorizing a payment with Apple Pay is being used with Face ID, the system looks for a score of 0.9. The accuracy of Face ID will only get better as technology develops, and although it is still pretty new now, later on bugs such as identifying identical twins won’t be an issue. The system is already very smart, considering it can detect people even when they are wearing sunglasses or a hat. The major variables of how accurate Face ID functions include pose, illumination, and expression (PIE). These factors influence how accurate the verification image will be, to compare to the enrollment image. The Face ID algorithm continues to update and adapt based on its surroundings when your appearance changes. A type of technology called the “Neural Engine” is an artificial intelligence network that acts like the brain, in which neurons continue to grow and learn, and this is what the Face ID algorithm is based upon. All in all, Apple calls Face ID “A revolution in recognition”. Just how advanced will this system get?
Comments