Home > News > Industry News > Facial Recognition Is Everywhe.....
Browse Categories
Face Recognition Temperature(5)
New Arrival(26)
Hot Products(12)
Fingerprint Scanner(22)
Handheld Terminal Series(14)
POS Terminal(5)
Time Attendance Series(14)
Access Control Series(6)
Personal Security Protection(4)
Sales Promotion(35)
Door Lock Series(11)
Camera Series(11)
Explosion Models
Contact us
Tel:+86-23-67305242 :
Facebook:Huifan Technology
Contact Now


Facial Recognition Is Everywhere. Here’s What We Can Do About It

  • Author:HFSecurity
  • Release on:2020-12-01

Facial recognition is a software that can draw, analyze and then confirm the identity of faces in photos or videos. It is one of the most powerful surveillance tools ever. Although many people only interact with facial recognition by unlocking their phones or sorting photos, how companies and governments use facial recognition will have a greater impact on people's lives.

If it is the equipment you own or the software you use, you can choose to opt out or turn off facial recognition, but the ubiquitous camera makes this technology more and more difficult to avoid in public. This widespread concern, exacerbated by evidence of ethnicity and the identity of the protester, has caused major companies including Amazon, IBM, and Microsoft to suspend sales of their software to law enforcement agencies. But as the ban expires and the technology behind facial recognition becomes cheaper, society will need to answer major questions about how to regulate facial recognition, as well as small questions about which services each of us is willing to use and which privacy to sacrifice everyone Are willing to do it.

How facial recognition software works

Facial Recognition Device
For decades, most people have seen facial recognition (video) used in movies, but they rarely portray facial recognition correctly. Each facial recognition system works differently (usually based on proprietary algorithms), but you can divide the process into three basic technical types:

Detection is the process of finding faces in an image. If you have ever used a camera that detects human faces and draws a frame around them for autofocus, then this technique can work. On its own, this is not harmful-face detection only focuses on finding the face, not the identity behind it.

Analysis (also known as attribution) is the step of drawing faces (usually by measuring the distance between the eyes, the shape of the chin, the distance between the nose and the mouth), and then converting it into a string of numbers or points, usually called "facial expression". Goofy Instagram or Snapchat filters use similar technology (video). Although the analysis may fail, especially when it involves misidentification, the problem usually only arises when facial fingerprints are added to the recognition database.

Recognition is an attempt to confirm the identity of a person in a photo. This process is used for authentication (for example, authentication in the security features of newer smartphones) or for authentication that tries to answer the question "Who in this picture?" This is where technology enters the more creepy side of things.

The detection phase of facial recognition starts with the algorithm for learning facial expressions. Usually, the creator of the algorithm does this by "training" the algorithm with facial photos. If you fill in enough pictures to train the algorithm, it will understand the difference between a wall socket and a face over time. Add another algorithm for analysis and another algorithm for recognition, and you will get a recognition system.

In the analysis and identification steps, the variety of photos sent to the system has a profound impact on its accuracy. For example, if the sample set mainly includes white people (as in the training of early facial recognition systems), the program will have difficulty accurately identifying BIPOC faces and women. In recent years, the best facial recognition software has begun to correct this, but the frequency of false matches (PDF) for white men is still lower than that of other populations; some software incorrectly recognizes that certain blacks and Asians are white. 100 times. Mutale Nkonde, a researcher at the Digital Civil Society Laboratory at Stanford University and a member of the TikTok Content Advisory Committee, pointed out that even if the system works well, gender identification problems still exist: "Tags are usually binary: male, female. This type of system cannot view non-binary or Even people who have transitioned."
Facial Recognition Device
Once the company has trained its software to detect and recognize faces, the software can look up and compare with other faces in the database. This is the identification step, in which the software will access a database of photos and cross-references to try to identify people based on photos from various sources (from portrait photos to photos grabbed from social networks). The results are then displayed, and the results are usually ranked by accuracy. These systems sound complicated, but with certain technical skills, you can build a facial recognition system yourself using off-the-shelf software.

A brief history of facial recognition

The roots of facial recognition can be traced back to the 1960s, when Woodrow Wilson Bledsoe (Woodrow Wilson Bledsoe) developed a measurement system to classify facial photos. Then you can compare a new unknown face with the data points of the previously entered photo. According to modern standards, the system does not run fast, but the idea has proven to be very valuable. By 1967, interest from law enforcement agencies had begun to rise, and these organizations seemed to have funded Bledsoe's ongoing research (never published) into a matching plan.

In 2001, law enforcement officers performed facial recognition of people on the "Super Bowl XXXV".

Throughout the 70s, 80s and 90s, new methods with easy-to-remember names such as "Eigenface Method" (PDF) and "Fisherfaces" improved the technology's ability to locate faces and recognize features, paving the way for modern automated systems Up the road. .
Facial Recognition Device
The dramatic transformation of facial recognition technology to public for the first time in the United States also caused its first major controversy. In 2001, law enforcement officers performed facial recognition of people on the "Super Bowl XXXV". Critics claim that it violated the rights of the Fourth Amendment against unreasonable searches and seizures. That year also witnessed the first widespread use of this technology by the police. The technology is operated by the Pinhelas County Sheriff’s Office, and the database is now one of the largest local databases in the country.

After a few years, the Illinois Biometric Information Privacy Act took effect in 2008, becoming the first law of its kind in the United States to regulate the illegal collection and storage of biometric information (including facial photos). Jennifer Lynch, director of surveillance litigation at the Electronic Frontier Foundation, described BIPA as a model for business regulation. She said: "Illinois needs to collect any biometric information notice and written choice consent." "At this point, Illinois is the only state that needs to do so."

The modern era of facial recognition began in the 2010s, as computers were finally powerful enough to train the neural networks needed to make facial recognition a standard function. In 2011, the face recognition function could confirm the identity of Osama bin Laden. In 2014, Facebook publicly disclosed its DeepFace photo tagging software. In the same year, facial recognition played a key role in convicting Chicago thieves. In the same year, Edward Snowden issued a statement indicating that the US government collected images to construct images. Degree of documentation. database. In 2015, the Baltimore police used facial recognition technology to identify protesters that occurred after Freddie Mac was killed by injuries to the spine of a police car.
Facial Recognition Device
Clearview AI became news in early 2020, when the New York Times disclosed that the company regularly compared its identification software with photo databases obtained from sources such as social media, news sites and employment sites on the Internet.

The facial recognition function was first introduced to personal devices in 2015 through Windows Hello and Android’s Trusted Face as a security feature, and then in 2017 the iPhone X and Face ID were introduced.

Since then, things have accelerated:

In 2017, President Donald Trump issued an executive order to expedite the use of facial recognition at the U.S. border (since private airlines have made their own efforts to integrate the technology).
In 2018, Taylor Swift's security team used facial recognition technology to identify stalkers, and China quickly increased its usage. Facial recognition technology is a general security measure in Madison Square Garden (Madison Square Garden), and retailers in the United States try to use this technology to track legitimate shoppers and shoplifters.
In 2019, a landlord in New York tried to install it to replace the key, and several schools tried the same method.
Today, a few cities-San Francisco, Oakland and Berkeley in California, and Boston and Somerville in Massachusetts-have banned government agencies from using facial recognition. The country also found the first false positive case, leading to arrest in the United States. After the brutal protests by the "Black Lives Matter" police began in June, several large facial recognition vendors including Amazon, IBM and Microsoft stopped selling their technology to law enforcement.

However, other new players have also entered the arena. Clearview AI became news in early 2020, when the New York Times disclosed that the company regularly matched its identification software with photo databases obtained from social media, news sites and employment sites (including Wirecutter and many others) on the Internet. Confirmation through testing-in the process of identifying suspicious objects. In May 2020, the American Civil Liberties Association (ACLU) announced a lawsuit against Clearview AI in an Illinois court, accusing it of infringing the privacy rights of Illinois residents under BIPA.Clearview AI is only an outlier if it is publicly censored: the existence rate of ethical software companies is also much lower, and these companies sell their software to local law enforcement agencies, usually not on the source or recognition algorithm Supervise or publicly review work.

Arguments for and against facial recognition

Facial Recognition Device
People who support facial recognition find the software very useful because it can not only identify criminal suspects, but also monitor known criminals and help identify abused children. In the crowd, it can monitor suspects in large-scale incidents and improve security at airports or border crossings. The longest-lived facial recognition software runs photos through government-controlled databases, such as the FBI's database of more than 400 million photos, including some state driving licenses, to identify suspects. The local police department uses various facial recognition software usually purchased from private companies.

Facial recognition can provide many benefits outside of law enforcement and can add convenience or security to everyday things and experiences. Facial recognition helps organize photos, helps secure devices such as laptops and phones, and helps blind and low-vision communities. This is a safer choice for entering business premises, conducting fraud protection at ATMs, incident registration or logging in to online accounts. Advertising and commercial applications of facial recognition are expected to bring various assumed benefits, including tracking customer behavior in stores to personalize ads online.

In an interview, Brenda Leong, a senior AI consultant and director of ethics and ethics at the Future Privacy Forum, suggested that supporters pointed out that facial recognition could replace loyalty programs or closed visits: “You just need to walk through a whole set of cameras, Things proceed very smoothly, such as stadiums, event venues, amusement parks, all of which are in use or have similar usage ideas."

Facial recognition helps to organize photos, helps secure devices such as laptops and phones, and helps blind and low-vision communities.

Opponents believe that these benefits are not worth the privacy risk, nor do they trust the system or the people who run them. The first point of contention is the collection itself-it is easy for law enforcement officers to collect photos, but it is almost impossible for the public to avoid taking photos. For example, the mug shot occurred before being convicted after being arrested. The recognition error rate is also problematic, from the false positive sense (identifying the innocent incorrectly) to the false negative sense (not identifying the guilty person).

The facial recognition software currently used by law enforcement agencies cannot yet be publicly audited, and the algorithms that support the detection and recognition software are usually closed proprietary systems that researchers cannot investigate. When the public does not know how these facial recognition systems work or how accurate, the public does not know whether these systems are used correctly, especially in law enforcement. Joseph Flores (Joseph Flores) is a software developer who uses machine learning for art projects in his spare time (Open: I have worked on related art projects with Flores for entertainment purposes , Not for profit), explained to me how he often deliberately biased the data set in order to produce the results he wants, law enforcement officers can also do something: "You can perform the same operation on the facial recognition data of law enforcement officers to Make sure that your friends cannot be identified and your enemies are mistaken for criminals." Flores added: "It is difficult to challenge the legitimacy or mathematical reliability that you cannot review. Especially when discussing data size. Uncensored, everything All are hypocritical, just modern physiognomy."
Facial Recognition Device
The public does not know whether these facial recognition systems are being used correctly, especially in law enforcement.

Another growing problem is that law enforcement officials are interested in real-time identification in real-time video sources or police camera footage. However, even cities that have enthusiastically adopted this technology, such as Orlando, Florida (where the police department also uses Amazon’s Rekognition software to try to identify criminal suspects from video streams in real time), the technology has failed to reach After the expected level, these efforts were withdrawn. expect. However, just because real-time facial recognition still suffers large-scale hits in real-time testing doesn’t mean it won’t be popular in the future. This idea is so shocking for some communities that California, Oregon, and New Hampshire have temporarily banned the practice.

The future of facial recognition and conditioning

Generally speaking, the future of facial recognition can take any of the following three possible forms: no regulation at all, and some regulation and prohibition.

No regulations

The "Black Mirror" episode depicts a world lacking facial recognition rules. Brenda Leong provides some examples: “It’s very easy to create very Orwellian futures, because the camera is everywhere, so you can see everything wherever you go. If you’re a student, then it’s actually It may be observing whether you are focusing on work or daydreaming. If you are an employee, please monitor your participation on your computer or tell you if you are wandering elsewhere. "The list of surveillance possibilities is almost endless Endless, such as China’s “Social Credit Score” or the London Police Force’s real-time use of facial recognition cameras, can glimpse a particularly harsh reality.


At the time of writing, there is a proposed U.S. federal-level law that prohibits the police and the FBI from using facial recognition technology, and another allows for exceptions through warrants. There is also a bill requiring companies to obtain consent before using facial recognition software publicly, and another bill prohibits the use of facial recognition software in public places. Although face recognition will definitely take some time, it is still unclear which of these bills (if any) will receive enough support to become law.

When anyone talks about regulating facial recognition, it needs to be divided into two parts: regulating commercial use and regulating government use, including law enforcement.

Leong emphasized that for commercial use, the main regulatory focus of any commercial function (loyalty program, theme park VIP visit or other) should be agreed. She said that facial recognition "will never become the default." "It should never be part of the standard terms of service or privacy policy. And never have to opt-out like what happens." The easiest way to understand how such regulations actually operate at the federal level is to look at Illinois State BIPA, BIPA needs to obtain the consent of the entity to collect and use biometric data (including facial fingerprints), and put forward requirements for the storage of data.

The list of surveillance possibilities is almost endless

However, consent can be tricky. It’s one thing for a store to ask you if you want to skip the question of entering an ID. It’s another for a store to use this technology to track shoplifters at all of its franchise locations. For example, Jennifer Lynch of EFF pointed out the case of a recent business district in London where the company placed a camera in a private business area where a working person passed by: “You can see the business district and say ,'Oh, well, we posted a slogan," Lynch said. "So people know that when they walk in the area or when their faces are recorded and captured, I really don’t believe people can really meaningfully agree in this situation. If you work in the area, you may not No choice.

Regarding the government's use of facial recognition, the recommended policy methods vary. Leong said that although the main focus of the "Future Privacy Forum" is the commercial use of facial recognition, the organization also wants to see regulation of government use. She said: "We would very much like to see public, intentional regulatory guidance on how and how the government should use facial recognition, even if it is really clear what level of warrant or possible reason the agency needs. Visit it."
Facial Recognition Device
Other groups, including EFF, believe that the supervision of law enforcement is far from enough.

Huifan Technology FacebookHuifan technology LinkedinHuifan Technology FacebookHuifan Technology Twitter

Related News:

Face recognition visitor machine makes front desk visitors smarter

The future development of face recognition will break through the scope of security applications

Website: www.hfteco.com

Website: www.china-attendance.com

Email: info@hfcctv.com