According to foreign media reports, the barber agency of New York State recently passed a new law to prohibit the use of face recognition and other biometric technologies in schools until 2022. Prior to this, San Francisco, Somerville, Massachusetts and other places have formally passed the law to ban face recognition software in public places.
Because of the accuracy of the current face recognition systems and the bias problems, the public are worried that these systems will be used to monitor and establish databases of sensitive face information, which will affect the local security.
Face recognition has always been one of the most controversial technologies. As an important personal identification technology, it is widely used in security, finance, transportation and other fields. However, due to the lack of data and single type of data, the recognition algorithm is more likely to make mistakes in dark skin, children and other people, which is regarded as “algorithm discrimination”. AI technology is difficult to rule out the small probability problem, so large technology companies have been repeatedly criticized.
Vulnerability of face recognition into security system
However, face recognition has entered into all aspects of life: pick up the mobile phone to sweep the face to pay the bill, complete attendance, check in the hotel and so on, which greatly facilitates life. While enjoying the convenience brought by technology, people should also deal with its potential risks.
As one of the important technologies of intelligent security, face image plays a similar role of “password” in application scenarios. However, the password can be “cracked”, and the face recognition system will be cheated and attacked. In 2017, CCTV’s “3.15” party showed face recognition technology and exposed its loopholes, which caused people’s panic about the technology. In 2019, a large-scale data leakage event occurred in a domestic professional face recognition enterprise, which pushed face recognition technology to the air of public opinion again. Although the relevant enterprises continue to study to improve the accuracy of biometric recognition, there are still some risks in this technology, not only facing the risk of information leakage, but also some legal hidden dangers, which may cause damage to the safety of national and personal life and property.
And different from foreign countries, Chinese enterprises in the application priority environment, face recognition and other new technologies are still growing savagely. According to the head of a cloud service provider, China’s regulation is very strong, but most of it is after the event. Therefore, before a vicious incident is exposed, the data rights of users depend on the moral level of the enterprise. “There’s no point arguing about where to do face recognition, because it’s all over the place.” How to ensure the security of face recognition in security system “depends on enterprise consciousness”.
Improvement and upgrading in progress
At present, many enterprises around the world have been involved in the research of face recognition technology and are committed to improving the accuracy of recognition technology. The difference is that each enterprise’s technology has its unique advantages. Through the continuous transformation of research results, the improvement and upgrading of security system based on face recognition can be realized.
The main purpose of living detection is to identify and judge the physiological information on the living body. It takes the physiological information as the life characteristics, so as to distinguish the biological characteristics forged by non living materials such as paper photos, electronic product display screens, silicone masks, stereoscopic portraits, etc. Living detection focuses on blinking and mouth opening and closing judgment, so as to better distinguish whether the face in front of the camera is real or fake, so as to improve the accuracy of face recognition technology. At present, the mainstream living detection solutions are cooperative and non cooperative living. The cooperative living body needs the user to make corresponding actions according to the prompt to complete the discrimination, while the non cooperative living body can directly detect the living body without the user’s feeling, which has better user experience.
◆ 3D face recognition
Since the face detection and recognition process is easily affected by ambient light, the active light scheme is adopted, which can reduce the influence of ambient light changes, thus improving the accuracy of face recognition. In addition, the traditional 2D face recognition can not record the depth information of the face, and can not present the complete face data information, so the silicone mask and false photos of the face can be taken advantage of. In contrast, when 3D sensor camera is used for face recognition, the built-in dot matrix projector can simultaneously project more than 30000 infrared points to the user’s face, which is difficult to observe with the naked eye, and its data is extremely rich, especially in the aspects of color, texture, depth, etc. After fully obtaining the face information, it can resist the attack of foreign video and photo, and improve the safety factor of face recognition technology. With the official operation of 3D face painting, shopping and other projects, more security products will apply this technology in the future.
Conclusion: privacy security is not only a technical problem, but also an application problem. In the future, it is still necessary for the government, the security industry and enterprises to make concerted efforts to balance privacy, security and convenience, and to clarify the boundary of face recognition, so as to make the application of face recognition technology more “secure” and “sense of gain”.
Editor in charge: Tzh