Article from WeChat public account:CV Intelligence (ID:CVAI2019), author: Yu Yang Yang, editor: Zhang Lijuan, cover: Oriental IC
Regarding face recognition, what problems are worthy of vigilance, and what are the problems that are just excessive anxiety?
A visitor brought a zoo to court. For no other reason, the newly deployed face recognition authentication system for the visitor’s dissatisfaction with the zoo will be used to collect personal data.
This is the first time in China that someone has brought a commercial organization to court because of the security and privacy issues involved in face recognition. It has once again brought this controversial technology since its application to the forefront.
Brushing security, paying, boarding, taking the subway bus, even getting a marriage certificate, adopting a pension… Face recognition is replacing keys, bus cards, bank cards and even ID cards, quickly infiltrating people’s clothing, food and shelter .
If you are willing to accept it, we are quickly entering an era of “people face brushing”. As a result, people’s anxiety and fear of face recognition are increasing.
The first thing that matters is the security issue. Can the face payment be easily cracked?
Secondly, it is a well-known privacy issue. It is also the most worrying issue for tourists who have brought the zoo to court. Does the zoo have the right to collect their own face data? What if the zoo leaks out its own data?
More controversially, in classrooms, construction sites, offices, etc., face recognition and even more sophisticated technologies were once used to “monitor” students and social animals.
This technology has caused people to be frightened, and the anxiety in sci-fi movies has doubled this concern.
But what about the face recognition, what are the issues that are worthy of vigilance, and what are the problems that are just excessive anxiety? CV Intelligence interviewed many AI company face recognition experts, algorithm engineers, face data collectors, related technology geeks and scientists, trying to open up the fog and controversy, and restore the most realistic face recognition technology as objectively as possible. Commercial condition.
Is your face safe?
A group of pupils surrounded a Fengchao express cabinet, and a young boy completed the face recognition with a printed photo of the hand in front of the courier.
A group of primary school students from Zhejiang First School accidentally discovered that they could use a printed photo to replace the real person’s face, deceive the Fengchao smart cabinet in the community, and take out the parents’ shipments. A photo can be “blind”, and face recognition has been easily cracked by a group of primary school students.
The skepticism about the security of brush face payment followed. If someone can easily unlock all the devices equipped with face detection with just one photo, can someone else easily steal my belongings?
In fact, the courier incident is just an extreme example of not typical and representative.
“A photo can be blacked into the system. It can only be said that this self-receiving system does not have a living recognition technology. It is estimated that the product manager feels that the express delivery is not worth a few dollars, and valuables will not be placed in the cabinet. I don’t think it’s necessary to have this feature.” An algorithmic engineer who developed a face recognition application explained to CV Intelligence.
But the item is not expensive, you don’t need live recognition technology. Could the product manager look down on us poor?
Of course not. “After all, living recognition technology is costly to develop, and the cost of collecting data can reach hundreds of thousands a day.” The engineer continued to explain to CV.
Cloud from a technology face recognition application developer also told CV intellectuals, this thing is actually the “living detection technology” in face recognition technology (Determine whether it is a real person, not a photo, video or mask, etc.) The security level is low, and it is necessary to add a high security level of living detection technology to this business.
The current face recognition is divided into 2D and 3D from the technical solution. Typical 2D in vivo recognition technology applications such as Alipay, you need to blink before the face verification, which ensures a completely still face, such as photos, can not be unlocked.
3D means that the system can get the depth information of the verifier, and how is the 3D in vivo technology safe? It is no exaggeration to say that the accuracy can easily identify twins or multiple births.
Now, whether it is a well-known big factory such as Alibaba, Tencent AI Lab, AI companies such as Shang Tang, Vision, Yitu, Yun Cong, the face recognition services provided by them are based on 3D living technology. 3D recognition in major exams and access control security systems has already played an unexpected advantage. For the exam, the exam, and the security? almost impossible.
How difficult is it to deceive face recognition?
Two GeekPwn2019 contestants and young geeks told CV that if they wanted AI to identify Huang Jianxiang as Ivanka, they managed to complete the challenge through image confrontation. “Three models, we only broke one, and the effect of the attack is not particularly good. It (AI) will still give you feedback, although it I think the picture is Ivanka, but the confidence is very low.”
Genius is still difficult to achieve 100% face recognition, it is not necessary to say that ordinary people who do not understand technology, it is almost impossible to break the system at the terminal.
Despise a face of technologyIdentifying application developers tell CV Intelligence: “Is it easy to be compromised? This not only tests the strength of each algorithm, but more importantly, it is resistant to attacks, such as photos, videos, masks, etc.. Relatively speaking, Face recognition is currently safe in the field of biometrics. Of course, it is not foolproof. To achieve face recognition security and scale, it requires technical level, laws and regulations and industry standards to be gradually improved.
For how to solve the security risks of brush face payment, cloud from a technology face recognition application developer told CV intellectuals, you need to pay attention to two aspects: first need to brush face + payment password as Payment transactions, to avoid simply paying for the payment business; Second, face information and personally identifiable information need to be saved separately to avoid the risk of simultaneous disclosure.
Of course, security is relative. No technology can guarantee 100% security. The commonly used 3D live face recognition is already a security method with high security and great difficulty.
Do face recognition deprive us of our privacy?
Painting face attendance, brushing face check, brush face payment, brush face registration… The ubiquitous camera captures the age, race, expression, identity, occupation, income, preference and personality of people all the time. The technology companies and related organizations that control the camera behind us will have an overview of our personal data and privacy.
The problem followed. People are beginning to worry that if the data is leaked or hacked by hackers, it can lead to the use of possible harm by criminals.
Face recognition technology’s powerful data collection and analysis capabilities have become a double-edged sword. On the one hand, it facilitates social management, on the other hand, it also makes people feel that their personal privacy is greatly offended.
For the collection and use of data, Defaced Technology’s face recognition application developer told CV Intelligence: “Is it easy to obtain, this includes obtaining face data and bottom database data at the business site, and having only any single data. It is impossible to complete the identification comparison. At present, any aspect of the face recognition business application scenario, such as data collection, invocation or comparison, needs to be carried out with the knowledge and consent of the user.”
Data labeling service provider Testin CTO Chen Guancheng also told CV that they the data collected must be legal.
“There are two aspects to the data collection annotation. One is security and the other is privacy. Security, all of our customers, almost all of them will ask that this data can only be given to me. Can not give other homes, so we will pass the laws and regulations, contract terms to guarantee our data, whether it is the entire data processing, or the security of my final data will guarantee the interests of partners.”
< /blockquote>
“Privacy, probably involving the data collected by citizens, we will require personal authorization. The collected individual citizens actually authorize us to use this data. This is what we guarantee from this perspective.” Chen Guancheng to CV It is explained that the commercial face recognition equipment we are currently in contact with, whether it is used for training data or collected data, is obtained through legal and formal channels.
As for data storage, many face recognition algorithm experts told CV that the core database of face biological samples is controlled by the core institutions such as the public security and central bank, and not the general commercial operators can obtain.
Face recognition devices installed on campus and work environments have caused controversy because they monitor people’s behavior and even their mental state. Face recognition “seeing through” people’s personal information and even psychological state does make people feel anxious.
The Vision Technology has moved the face recognition device into the classroom, and the storms used to analyze the state, expression, and concentration of students have not completely ended. A wave of unrest, a wave of up again. In the United States, a company called Gaggle is monitoring students in the school in much the same way.
As issues get noticed, schools are beginning to regulate the use of such devices. Commercial companies are beginning to reflect on the privacy and ethical issues in face recognition applications.
Cloud from a technology face recognition application developer told CV intellectuals, using face recognition technology, whether it is classroom management or teaching effect analysis, it is inevitable to collect massive data. Personalized tracking of student learning behavior, whileThe goal is to generate personalized learning programs through analysis to help students improve their learning efficiency. In this ideal situation, it is in line with the future trend of education.
In the developer’s view, face recognition as a technology does not have any problems in itself. The key is how to use it. The same face recognition technology, which helps kindergartens and primary schools to establish a smart security system to find lost young children or guard the elderly, is rarely controversial.
“The use of AI technology in educational scenarios requires further advancement of the rule of law, the establishment of relevant laws and regulations, and compliance with laws. This is to ensure that technology is not abused, restrict the negative impact of technology, and truly play. Its key to promoting the role of education and teaching.”
There is widespread controversy with the application of face recognition technology. In order to promote the use of AI, contempt for the introduction of the “Artificial Ethics Code of Ethics” based on the company’s own management standards in July 2019, from the basis of legitimacy, human supervision, technical reliability and security, fairness and diversity, responsibility Traceable, data privacy protection six dimensions, to clearly define the correct and orderly development of artificial intelligence.
How to use technology, different societies have different choices, but the unchanging consensus is that technology should always be used properly and correctly.
The American Academy of Arts and Sciences and cancer immunologist Carl June told CV that for the entire society, including scientists, we have a responsibility to think about how these new technologies should be used, and now technology development. Very fast, but different societies have different enthusiasm and willingness to accept and adopt new technologies.
“My responsibility as a scientist is to ensure that cancer can eventually be cured. I hope that this can be achieved through advances in technology, but the challenge for society as a whole requires everyone to work together.” Carl June also believes that How to make good use of technology and cope with the privacy and moral hazards brought about by the abuse of technology is the common responsibility of the whole society.The power cannot be resolved.
from crazy to rational
After several years of intensive and rapid exploding, face recognition began to enter a stage of rational development. The problems caused by security, privacy, ethics, and indiscriminate over-application are flooding, and reflections follow.
With the “China Face Recognition First Case” filed, people began to pick up legal weapons to deal with the concerns raised by the excessive application of face recognition. Whether it’s a school or a zoo, more organizations that used face recognition applications have begun to consider the side effects that can be created by applying new technologies while giving them management convenience.
Scene, Cloud, and other companies that provide face recognition services are repeatedly considering the security of technology and developing relevant ethical standards to regulate themselves. Technology and business geniuses are thinking about how to win market competition, how to maximize profits, and how to hand over a good-looking financial statement, while considering the safety, privacy and dignity of users as human beings.
Integrated interviews with multiple industry insiders, CV Intelligence found that the industry has a relatively mature solution for the security and data privacy issues that the public is concerned about. As for the use of face recognition to “monitor” students and social animals to master student learning and workers’ work, it is currently the most controversial area.
The problems brought about by new technologies cannot be ignored, and it is necessary to slow down the excessive anxiety caused by the truth.
Article from WeChat public account:CV Intelligence (ID:CVAI2019), author: Yu Yang Yang, editor: Zhang Lijuan