Capture the face micro-expression, analyze the person’s heart, face-depth application of face recognition or change the interpersonal relationship.
“brush face” to unlock the phone, withdrawals are small meaning is worrying is “brush face read heart”
these two days, brush face unlock the phone, brush face in ATM machine money The majority of the reader’s interest. In fact, many IT people predicted that this year will be “the first year of face recognition,” the development of this technology has far more than unlock the phone, money and other areas, it allows everyone to become “micro-expression” analysis experts , There may be subversive thousands of years the formation of interpersonal relationships. Purple cattle journalists interviewed the United States Rainbow company CTO, chief scientist Dr. Wang Jin, for your detailed understanding of the face of this “black technology.”
You can not think of the technical height
pores, wrinkles can not escape face recognition
Dr. Wang Jin Of the resume, his first face recognition application began 15 years ago, Samsung mobile phone camera system, when a small camera at the people, each person’s face will appear a small red frame, the processor in accordance with this A region of metering, to get the correct exposure parameters, this guarantee that you shoot a qualified portrait of the technology is the first face recognition. Wang Jin told reporters that his earliest design of the face recognition system in the human face to collect five data, eyes, mouth and nose, and now the software can capture the human face every pore, wrinkles, To help the phone to confirm whether you are the owner of this machine.
“The most important data to measure face recognition is the success rate, and now face recognition has begun to use in the public sector, such as the station’s pit stop system, public security monitoring system, etc., and the phone’s face Identify the underlying technology with the relevance, but the focus is different, “Wang said,” public security field of face recognition is more like a needle in a haystack, looking for the suspect’s work can be done through a long time to identify the match is now better Of the system can identify 1 million images in a second.Laser unlock the need for identification is more refined, the need to do the error rate is very low to bring a better user experience. “
You may have this The question?
These two days there have been a lot of face recognition on the pieces, the makeup can “brush face” it? Will it be asleep? … … Wang Jin one by one answer.
of the makeup, face recognition can recognize you?
“Face recognition has a learning mechanism, each use will store different light, the environment of the facial information, you can say that a machine will be more familiar with you, this change in the environment can even Time, the growth of age to change, not because you increase the number of wrinkles do not know you. “Wang Jin told reporters that the case of light makeup almost does not affect the face recognition, the machine collected information than people can Think of much more, the location of the pores, the thickness of wrinkles, the size of girls lying silkworm eyebrows and large and small information will become the basis for identification. “In addition to the two people themselves, like the same heavy makeup, in this extreme case, it may affect the accuracy.”
For sleep can unlock the phone, Security considerations, Apple’s FaceID requirements to unlock one of the conditions is to capture the user’s eyes. If you are asleep or close your eyes, it is impossible to complete the unlock or pay for the face alone. However, the general face recognition technology to identify the face in sleep, technically no difficulty.
wearing sunglasses, masked, can escape face recognition?
If a person covers the face, such as wearing a pair of sunglasses, or mask cover the second half face, or simply blindfolded face recognition technology can do something?
Wang Jin said: “If your friends and friends wearing sunglasses, masks you can recognize it? In fact, you still can recognize a great chance, and as long as you can do things, the computer must Can be done better. “
His team has also tried to experiment with large data analysis of human face bone, muscle, eye relative between the impact of millions of pictures through the calculation, The result is a surprise, face recognition can be in the case of incomplete facial information, like the same as the puzzle combination of the remaining information. According to the United States, “Popular Science” reported that the British Cambridge University and Indian scientists use depth learning technology and database identification masked, the accuracy rate of up to 69%. “If you only cover part of the face, it is difficult to hide the entire face structure, the artificial intelligence model is still able to predict what the hidden facial structure is like.”
that with a live photo muddle through, ok ok? In addition to the data algorithm, the hardware in the face recognition is another “door god”, Wang Jin introduced the first camera on the phone is a single lens camera, followed by a dual camera camera, Apple’s new Generation products used in the deep camera, will cast more than 30,000 invisible eyes, for the user’s face to draw accurate and meticulous 3D map, with photos and other things posing as pass.
“The single-lens camera can only take pictures of the plane, and the dual-lens camera is similar to the 3D movie we see.Through the camera can collect the depth of information, such as judging the distance between the nose and ears, the three-dimensional level of the face Image capture the most powerful camera. “Wang said.
“Face recognition is currently the biggest bottleneck is in vivo detection, simply to confirm that the lens is identified before is a real human.” Wang Jin told reporters that people are different from pictures, projection, video and All the means of posing has been the subject of continuous research in the field. Today’s algorithm has been able to filter out real photos such as “Li Gui”.
what about the future!
can determine the micro-expression, inner secret will be difficult to hide
facial facial expression, often the reaction of the inner world. Human use to cover up the inner real thoughts, although it may be used to cheat, but most of the time also contribute to the normal development of interpersonal relationships.
Since face recognition can identify different people, can not recognize the human face? Wang Jin told reporters that his team also tested a few years ago, people face the little action recognition is not difficult, as long as the database analysis, the future in front of the camera lens, some of the inner activities of people will not cover.
The Economist magazine says there is a far-reaching impact on the continuous recognition, recording and analysis of facial subtle expressions.
If the wife with face recognition technology, aware of each of the husband’s strong pressure yawn; boss with this technology, aware of the employee every trace of angry expression, marriage and workplace relations will become more real, but More likely because people can no longer hide the real idea, and thus more tense. The basis of social interaction may also change, rely on the development of interpersonal relationships may be subversive.
“anti-face recognition” technology is also progress
1 / anti-face recognition make-up
Germany Artist Adam Harvey developed anti-face recognition technology make-up, mainly blurred nose, eyes and face shape and other facial features. His project also has a camouflage-like headscarf, composed of many large and small face patterns, allowing facial recognition equipment disorders, interference with its algorithm.
Russian hackers have also developed a “anti-face recognition algorithm” to make-up way to draw a specific line in the face, in order to interfere with facial recognition technology. Japanese experts have developed a kind of glasses, a bit like a sleep goggles, is a mesh-like, does not affect the line of sight, but can reduce the nose and eye and other key issues.
2 / anti-face recognition glasses
Area of the reflective, thus interfering with digital face recognition software.
Previously they also developed a near infrared LED lights with anti-face recognition glasses, this light is not visible, in the camera is very bright, so that face recognition software can not play a role. Although this kind of glasses is very effective, but too cumbersome.