Special Reporter of the newspaper Ren Qibao reporter Ni Hao Liu Yang
On the morning of the 13th, Beijing time, Apple held its 2017 fall conference and unveiled a range of new mobile devices, including the long-awaited iPhone X, the 10th-anniversary special edition. The iPhone X broke away from the traditional naming system and immediately drew attention due to its Face ID facial recognition feature. Soon after its release, numerous discussions about the security of this new device flooded the domestic internet. Is the facial recognition technology used on iPhones reliable? What are the future possibilities for face recognition applications?
With the rise of mobile payments, many people now only need to carry their phones to complete most transactions. In the future, people may not even need to carry their phones as a "foreign object" — instead, they could simply use their face as a means of payment. Can your face really be used to go anywhere?
"Swiss Face" Unlocks the iPhone X
The iPhone X removed the Touch ID fingerprint sensor and replaced it with Face ID. All the sensors are located in the "notch" at the top of the screen. Using an infrared camera, Apple projects over 30,000 invisible light points to create a unique depth map of the user's face. When unlocking, this is compared with an infrared scan. According to the Wall Street Journal, Apple’s marketing director Phil Schiller stated that Face ID can adapt to changes in users’ appearances, such as when they wear glasses, change hairstyles, or put on hats. It works in both bright and dark environments.
The U.S. magazine Time reported that Face ID supports Apple Pay and all current third-party apps that use fingerprint authentication. However, during the event, Apple’s senior vice president of software engineering, Craig Federico, was temporarily unable to be recognized by the system, causing some embarrassment. Apple experts claim that facial recognition is highly secure because the chance of someone else having the same facial features is extremely low—only one in a million, compared to one in five thousand for fingerprints, unless you have an identical twin.
On the 13th, Dr. He Ran, a researcher at the State Key Laboratory of Pattern Recognition at the Chinese Academy of Sciences, explained to the Global Times how face recognition works: cameras capture face images, which are then scaled uniformly from different angles and distances. These images are then fed into a deep neural network to extract facial features, which are compared with data in a database. If the match exceeds a set threshold, the device recognizes the person.
What is the current level of "face scanning"?
Wang Yanhui, secretary-general of China Mobile, told the Global Times that according to Apple’s disclosure, its facial recognition technology is currently secure, but the real challenge lies in large-scale daily applications. Whether consumers accept the convenience of face recognition is the biggest test for Apple’s new system. “Authentication is a high-frequency task. Every consumer is verified hundreds of times a day. If the process is slow or inaccurate, users won’t adopt it.â€
How accurate is face recognition? Yang Fan, co-founder and vice president of Shangtang Technology, told the Global Times that accuracy varies depending on the application scenario, so it cannot be generalized. As technology advances, face recognition is expected to break through higher thresholds. Currently, it is more mature in areas like Internet finance and security systems.
Regarding countermeasures like disguises or masks, Shangtang’s algorithm can compare faces even if up to 40% of the face is covered. With continuous improvements in deep learning algorithms, face recognition can work under greater occlusion.
According to reports, face recognition has already been widely adopted in China, with high accuracy in security applications such as residential access control and company attendance systems. However, there are still technical challenges in certain specific scenarios. For example, during rush hours at the Beijing Subway, where dozens or even hundreds of people enter simultaneously, accurately and quickly identifying each face remains a challenge. Additionally, low-resolution scenes, night conditions, rain, snow, or haze, and high-speed movement also pose difficulties. Another key issue is ensuring that the system can distinguish between a real person and a photo or video.
Is the "face" phone safe?
According to a report in the U.S. publication "Linkage," a common weakness of biometric unlocking methods is that physiological characteristics like faces and fingerprints are difficult to change. Once compromised, it's much harder to fix than a password. The article noted that the safety of facial recognition depends on the context. For most users, it's relatively secure against theft since a thief would need to hold the phone up to the victim's face to unlock it. However, for those in high-risk situations, being forced to unlock the phone is a concern. Apple has addressed this by requiring users to open their eyes and look directly at the screen to unlock.
Matt Green, an information security expert at Johns Hopkins University, said, “I'm worried that Apple hasn't fully disclosed the technology behind Face ID. Existing 3D sensors can solve issues like photo attacks, but Apple hasn't shared details. So we don't know how secure our phones truly are.â€
Yang Fan believes that face phishing attacks can be effectively prevented using technical solutions. Current industry defenses include live detection technology, which identifies media properties and distinguishes between synthetic and natural images. Shangtang Technology can detect high-definition photos, PS images, 3D models, and face-lifts. According to the company, the success rate of such attacks is roughly two-thirds, and current defense mechanisms can block 95% or more of them.
How big is the future of paying with "face scanning"?
Many industry experts are optimistic about the future of face recognition technology. Yang Fan believes that as a non-contact biometric method, it offers great convenience and will be applied in more scenarios beyond just phone unlocking. For instance, some financial services allow remote identity verification via a phone scan.
Yan Shuicheng, dean of the 360 Institute of Artificial Intelligence, told the Global Times that face recognition isn't perfect in all situations. With the rapid development of technology, fake heads, holograms, and face tracking techniques are emerging, making single-factor face recognition limited and not absolutely secure. Therefore, for high-security applications like privacy and payments, it's better to combine face recognition with other biometrics like voice, fingerprints, or iris scans to enhance security. â–²
Paste Type Rotor,Dc Variable Frequency Motor Rotor,Dc Motor Rotor,Adhesive Dc Variable Frequency Rotor
Suzhou Mountain Industrial Control Equipment Co., Ltd , https://www.szmountain.com