万人炸金花

You are here: Kangba Media >> Financial Management >> View Article

alert! The "blind zone" behind face recognition

China Youth Daily January 12, 2020

While face recognition brings convenience, it also raises concerns. Relevant experts suggest that relevant legislation be strengthened to provide access scenarios and conditions for face recognition, clarify the qualifications of enterprises, and clarify what punishments should be accepted in the event of violations.

Brushing your face is quick, but there are many problems.

Face recognition technology began to land in many scenes, and corresponding disputes followed.

In October 2019, a consumer in Hangzhou requested the consumer to brush his face to enter the park, and brought the zoo to court, known as the "China's first case of face recognition."

Recently, Nandu Personal Information Protection Research Center released the "Face Recognition Landing Scene Observation Report", which unveiled the "blind zone" behind the current face recognition application in China.

"Swipe photos" can also open the door

In October 2019, members of the artificial intelligence ethics group of Nandu Personal Information Protection Research Center came to two public rental housing communities in Chaoyang District and Fengtai District of Beijing to conduct field tests on the community's face recognition system.

When the members of the research group tried to take photos of the residents of the community, they pointed the mobile phone photos at the face recognition machine downstairs, and a crisp "Door lock is open, please come in" came out of the machine. Then the door of the corridor opened, and the team members walked in easily.

Due to the low price, there has been a long-term violation of leases in public rental housing, which has always been a headache for managers. In January 2019, the Beijing Municipal Committee of Housing and Urban-Rural Development issued the "Notice on Further Strengthening the Supervision and Management of Public Rental Housing Sublease and Lending Behavior", which proposed that public rental housing projects included in the Beijing affordable housing construction plan should fully adopt face recognition and smart door locks And other technologies, and strengthened face recognition and other technical measures combined with access control to achieve that non-tenant family members are not allowed to enter the building unit door at will.

The team members selected several public rental housing communities for testing. They found that some community face recognition security technologies are not enough. "We are not actually using high-resolution photos," said Feng Qunxing, a member of the research team. At present, face recognition technology can be roughly divided into 2D and 3D recognition technologies. The former is imaged by a 2D camera and the latter is stereoscopically imaged by a 3D camera.

Generally speaking, 3D technology has high security but also high cost. Some experts said that what can be swiped with photos are basically 2D face recognition devices and ordinary cameras. In this case, you can avoid the attack of false photos by replacing them with infrared binocular cameras and adding improved algorithms.

In addition to security issues, Feng Qunxing and colleagues also found that some community's face recognition technology equipment was "unfriendly" to people with disabilities. At the gates of some neighborhoods, the camera height of the face recognition gate is about 1.2 ~ 1.6 meters, and ordinary people can pass normally, but it is difficult for people with disabilities in a wheelchair to reach the height of the camera. "These problems are also related to public rental housing in Beijing. The management department reflects that they attach great importance to the investigation and rectification of these issues. "

Out of the community, the team members also walked into campuses, shopping malls, public toilets and other places to experience the convenience and security of face recognition equipment.

In Beijing, members of the research group found that although face recognition has not been widely publicized on campus, its popularity is not low. Some schools use face recognition systems for access control, classroom attendance, and monitoring of course quality. Some students said that the current facial recognition system of the school classroom where they are located can detect the head-up rate and front-seat rate of the students to monitor the quality of the teacher's courses. But some students are concerned that such technology is suspected of violating their privacy.

The members of the research group found that in the education field, some educational institutions currently claim in the publicity materials that they can grasp the emotions of students through face recognition. When the members of the research team made an unannounced visit as a parent, a teacher from an educational institution sent a video. In the video, the back-office system could detect the mood of the students at any time and send out reminders such as "the child seems unhappy".

Members of the research team also visited two Beijing shopping malls using face recognition technology on the spot. One of the malls records customer consumption trajectory through face recognition. If the customer brought by the tour guide has consumption, the tour guide can get rebate. However, these did not obtain the consent of the customer, and several customers told the researcher that they did not know that they were brushed and their tracks were recorded.

A shopping mall in Xidan uses face recognition cameras at the entrance to count passenger traffic. Relevant staff said that the system only counts data and does not store customer photos. The members of the research group believe that the main problem of face recognition in the shopping mall scene is that customers are not fully informed and informed consent is obtained.

Nearly 80% of respondents are concerned about personal data being leaked

In addition to field investigations, the research team also conducted an online questionnaire, which mainly investigated the public's attitude towards face recognition, including whether it was more convenient and safer after using face recognition, and 6,154 valid samples were recovered.

The questionnaire data shows that more than half of the respondents have encountered problems with face recognition. Among them, the proportions in public rental housing, transportation, campus, shopping malls and other scenarios are 59.33%, 59.86%, 63.28%, 58.87% and 60.33%.

Some interviewees reported that factors such as hats, glasses, makeup, lighting, and angles will affect the accuracy of face recognition. Other interviewees said that there was no prompt on the face recognition display, and they did not know where to place their faces, and they could not grasp the distance from the camera.

In public rental housing, shopping malls, campuses and other scenarios, more than 60% of respondents believe that face recognition is more secure, but many respondents also expressed concern about the leakage of face recognition data. Among them, 79.31% of the respondents worry that after handing over facial data to the operator, the operator does not have sufficient ability to ensure that the data is not leaked. 65.17% of the respondents are concerned about the increase of online false information such as face-changing videos, and 49.57% of the respondents are worried that criminals use fake information to perform fraud or stealing.

In terms of transparency, nearly half of the respondents indicated that they did not sign the privacy policy or were not sure whether they signed it. More than 40% of the respondents did not know how their face data was stored. When asked if they would like to be able to view their stored information and have channels to delete, 83.37 respondents chose "yes", which represented an overwhelming proportion.

The Cyber Security Law clearly states that network operators should collect and use personal information in accordance with the principles of lawfulness, justification, and necessity, and openly collect and use rules, expressly the purpose, method, and scope of collecting and using information, and obtain the consent of the person being collected.

The members of the research group believe that face recognition technology has a fast landing speed and many scenes, but whether the application of face brushing is justified is worth considering. In the process of technology application, it is necessary to fully adopt public opinions to demonstrate the legitimacy and necessity of the application. In system design, more humane considerations should be introduced. "Face recognition is a new technology. When face recognition changes from a trend to infrastructure, it should take into account the vulnerable groups in society, rather than cause new social injustice."

The report recommends that government departments strengthen relevant legislation, stipulate the access scenarios and conditions for face recognition, clarify the qualifications of enterprises, and clarify what punishments should be accepted in the event of violations.

At the beginning of 2019, the four departments of the Central Cyberspace Office, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the General Administration of Market Regulation jointly issued the "Announcement on the Implementation of the Special Governance on the Collection and Use of Personal Information by App Illegal and Illegal", and organized the nationwide collection of App Illegal and Illegal in 2019 The special use of personal information governance has opened a strong prelude to the collection and use of personal information.

Regarding the supervision of face recognition technology, Hong Yanqing, deputy leader of the App Special Governance Working Group, believes that it is still necessary to distinguish between different uses of the technology before considering different legal frameworks to regulate it. In his work, Hong Yanqing found that currently there are at least six uses of face recognition technology, counting, identification, authentication, monitoring, forgery and snooping. In terms of counting, enterprises' face recognition technology is usually used to calculate the number of uses of a product, and does not involve authentication. For this purpose, do you need face features to achieve the purpose of counting? Hong Yanqing feels that it is necessary to further Inquired, "Should you use a bull's knife to kill chickens?"

For identification and authentication purposes, face recognition technology generally identifies who a person is, and then pushes some information to the merchant through authentication or even matching with a database. Hong Yanqing believes that in this case, a personal information protection framework can be used for regulation. Such behaviors as video face-changing and falsification can be regulated in accordance with the relevant laws and regulations on portrait rights. For snooping on such behaviors, they should be regulated in the relevant categories of personality rights and privacy rights.


  • Previous post: China's economic stability and progress
  • Next post: Don't Let Prepaid Cards Be Like "Overdraft" Consumer Trust