So the question is: What are the reasons for the ban or suspension of the use of this technology in more and more cities in the United States?
Defective face recognition technology
Objectively speaking, the face recognition technology in the United States has always been at the forefront of technology and has been operating at the commercial level. Amazon launched the face recognition software Rekognition as early as 2016. In addition, similar products include Google Cloud Vision API, IBM Watson Visual Recognition andMicrosoftFace API.
Despite this, current face recognition technology is not 100% accurate. In 2018, an article published in the New York Times English website pointed out that today's very popular AI application face recognition has a huge difference in accuracy for different races. Among them, the error rate for black women is as high as 21% -35%, while the error rate for white men is less than 1% - which is very politically incorrect in the United States.
Another Oolong event worth mentioning is: Giants Amazon launched the image recognition AI system “Rekognition” in 2016. It has identified 28 US congressmen as criminals, which has made American society feel awkward and full of face recognition technology. question.
Dr. Zhang Xiaobo from the Institute of Automation, Chinese Academy of Sciences has said that lighting, posture, decoration, etc. all have an impact on robot face recognition. For face image acquisition in non-cooperative situations, the occlusion problem is very serious.
Especially in the monitoring environment, the monitored object may wear glasses, hats, etc.AccessoriesThe captured face image may be incomplete, affecting subsequent feature extraction and recognition, and even causing the face detection algorithm to be invalid.
Although the size of the face recognition algorithm decreases with the increase of the size of the face database, how to maintain or improve the recognition rate of the face recognition algorithm in a large-scale application environment is also a very important issue.
Who is your face?
In terms of security, face recognition technology is also widely questioned in the United States.
For example, the information storage for face recognition is still based on a computer-aware language, ie a number or a specific code. As the value of these data increases, the risk of hacking will increase.
In other words, the support behind human face recognition technology is data. If these private data are stolen, your face may not only belong to you, but may belong to anyone.
At the same time, as mentioned earlier, face recognition technology has different accuracy rates for people of different races. Both the author's research and the manufacturer's experiments have clearly found that the female dark ethnicity recognition error rate is higher. Racial discrimination is a very sensitive topic in American society. If the relevant laws on face recognition technology have not been perfected, then face recognition technology will still be in dispute.
However, the bigger controversy is that American social groups are resentful and worried about “monitored” itself.
Earlier, Amazon and many other technology companies also worked with the US police to apply facial recognition technology to video surveillance and police-wearing camera lenses. Many US state governments, including Oregon, have become Rekognition users.
Amazon's face recognition software, Rekognition, recognizes, analyzes, and tracks in real time. It compares the information it collects with thousands of images in the database in a matter of seconds. Law enforcement agencies have applied this technology to the search for missing people and identifying suspects in terrorist attacks.
However, the American Civil Liberties Union (ACLU) has “written” the government: Amazon should stop providing the US government with a powerful and “dangerous” face recognition system Rekognition.
Shortly after the league’s “booking”, nearly 20 Amazon’s shareholder institutions sent letters to Jeff Bezos, raising concerns about the problems and implications of Amazon’s Rekognition being applied to law enforcement.
These shareholders wrote in the letter:
Amazon shareholders are concerned that such government surveillance facility technology may not only pose a privacy threat to national customers and other stakeholders, but may also increase the substantial risks faced by our company, negatively impact our company's share price, and increase financial for shareholders. risk.
Shareholders are also concerned that Rekognition may violate the civil rights of women, people of color, immigrants and other marginalized groups, who are particularly vulnerable to the police's focus.
Not only do Amazon shareholders believe that Rekognition will invade privacy, but the American public has a similar view. Applying face recognition technology to front-end video surveillance cameras may break the balance between privacy and usability.
Assuming that the US police has several such security cameras and a "blacklisted" photo library of suspicious people, then if anyone else has something similar to these suspicious people, once they enter the security camera of the police, it is possible Interrogated by police officers. And most Americans don't want to live in that world.
Not only is Amazon's Rekognition controversial, but US customs and border protection programs are also controversial because of the application of face recognition technology.
The program matches passengers through its cloud facial biometric system. This information matching service is designed to replace manual ticketing and to change the process of displaying paper documents on the road to automatic identification.
This facial recognition system is designed to be fast. Moreover, U.S. Customs and Border Protection (CBP) believes that facial recognition is more accurate than humans when comparing passport photos and faces. In 40 days, the system has helped find three imposters at Washington Dulles Airport.
But this technology can be used without your knowledge, and the immutable data collected by the facial recognition system, that is, facial information. This has caused people's concerns. If someone's information has been saved, then the system can keep track of the person's whereabouts.
Moreover, CBP has not implemented this opinion to the public and has implemented this technology at the airport, breaking through the boundaries of US law.
In the face of controversy, CBP stressed that the existence of this project is not for surveillance, and its implementation is in line with the law. US citizens can opt out of the program and their photos will not be retained for long periods of time. Not only that, but CBP also issued a statement proving that there are many laws that allow it to collect biometric information.
But privacy law experts point out that although the laws listed by CBP apply to non-citizens, they do not explicitly mention the situation involving citizens. Moreover, a report published by the Georgetown University Center for Privacy and Technology in 2017 stated that it has been evident in the past 14 years that US citizens have been absent from the relevant laws of the program.
CBP is still questioning, not to mention other US government agencies.
The acceptance and integration of technology by a social group is not necessarily smooth sailing; especially for countries like the United States that focus on political correctness and privacy first, the application of face recognition technology is facing an increasingly serious embarrassing situation. .
It is conceivable that face recognition technology may face more and more opposition in the United States, but whether this is a good thing or not is not known.