POS is being eaten again! Ma Yun "brush face payment" stunning Germany! Alipay ambition is exposed!

This new payment authentication technology was developed by Ant Financial and face++ financial to replace traditional passwords in the payment authentication stage after shopping. Alipay’s “empty payment” on April Fool’s Day seems to be not just a fool's day joke, getting rid of the password, relying on biometrics such as face to complete the identity authentication and payment life, it is indeed gradually getting closer.

Recognition accuracy is higher than the human eye

Is face recognition accurate? This is a lot of people's doubts, but what most people expect is that the accuracy of computer face recognition is already higher than the human eye. Tang Xiaoou, a professor at the Chinese University of Hong Kong, previously published a paper at cvpr2014, the top conference of computer vision international, saying that the accuracy of computer algorithms for recognizing faces has reached 99.15%, while the accuracy of visual recognition is about 97.52%. As the world's leading face recognition platform, face++financial face recognition technology achieves the world's highest accuracy of 99.5% in the lfw international public test set. At the same time, it uses "interactive instruction + continuity judgment + 3d judgment" Technology, its ability to face face recognition anti-counterfeiting attacks has also reached the world's leading level.

Inside the ant gold suit, there is also a team dedicated to biometrics, called “柒 workshop”. Zhang Jie, a senior technical expert of the team, said that the face recognition technology is based on neural networks, allowing computers to learn human brains. And through the extensive training of the "deep learning algorithm", it becomes extremely "smart" and able to "recognize people"; and the reason why the accuracy is higher than the human eye is that the computer can pay attention to more key details, and Through the algorithm, some interference factors are eliminated.

At the same time, human eye recognition is easily affected by the operator's own intelligence, experience, and seriousness. For example, on March 15 this year, CCTV exposed the problems of offline bank card processing, although the fake ID card bought by the reporter The photos are very different from themselves, but because the staff did not conduct a strict check, the reporter successfully executed three fully functional bank cards in the four banks. If you use technology to identify, such a fake ID card will be difficult to get away with.

Users can participate at zero cost

Interestingly, face recognition, a technique that sounds extremely cool, can be very "grounded" if applied to practice. Compared to many other biometric technologies such as fingerprints and irises, face recognition is easier to promote on a large scale because almost all users can participate at zero cost.

First of all, face recognition does not require the user to submit photos by himself. According to relevant regulations, a qualified institution can apply to the National Citizen ID Card Number Service Center when it needs face recognition, and collect the photos and the authority of the department. After the photo library is compared. That is to say, when the user performs face recognition, he only needs to open the camera of the mobile phone or the computer and shoot the face of his or her face. Today, when the smart phone has been fully popularized, the participation threshold is negligible.

Secondly, privacy issues that users are likely to worry about can be effectively avoided in the field of face recognition, because the source of photos is extremely authoritative. At the same time, a unique "desensitization" technology can blur the photos into unrecognizable eyes, only computers. The images can be identified to protect the privacy of the user.

Brushing face payment is not a dream

So, what kind of application space does face recognition technology have? Is it possible for everyone to pay for their faces in the future?

Ant Financial said that its enthusiasm for biometrics is mainly for the technical storage of Internet finance business, and the relevant specific products are currently being prepared.

Pi Dong, the head of the “柒 workshop”, said that in the past, we used to use the user name, password, and other things that users know, ID cards, and dynamic passwords to verify the user's identity. In the future, it is likely to be converted into users. Innate biometrics to identify identities such as fingerprints, palm prints, faces, irises, etc.

In the past, the world's top technology companies have been focusing on biometric identification. For example, Apple introduced fingerprint identification touch id. Samsung has also begun to test the water face to unlock. Alipay has also introduced fingerprint payment. The biometrics of the user who “sneaked away” is likely to replace the password and become the mainstream authentication method. In the future, many businesses that need to be physically present by the user may also need to “smear a face” in front of the computer or mobile phone. Quickly verify the "real body", no longer need to find outlets, queue time-consuming.

Although ant Jinfu has not given a clear answer to the specific application scene product form, it is not difficult to imagine that if face recognition is approved by various regulatory agencies, then Internet finance applications like brush face payment will have May become a reality.

In the field of mobile payments and internet finance, an era of “eat by face” is coming to us.

Supplement qa:

Question 1: Can God's make-up be recognized?

A: The computer recognizes that the above two are the same person, and the computer will pay attention to many key details that are not easily noticed by the human eye.

Question 2: The naked eye feels like it, can the computer recognize it?

A: The above picture is a picture of Miss Korean beauty pageant. It looks very close to the naked eye, but the computer can be successfully distinguished. The number of people an ordinary person can accurately identify is very limited, and it is usually accurate to identify acquaintances. It is easy to identify strangers, but the computer can accurately identify faces of tens of millions or even billions.

Question 3: The way to identify is to compare my own photo with my ID card. If my photo is very different from mine, I have changed a lot in recent years. Can face recognition technology be recognized?