Facial recognition technology is a new era in video analytics, video surveillance and access control systems. The most reliable method to measure your face. That is, the FSB can, but ordinary people cannot.

Modern integrated security systems are capable of solving problems of any complexity at all kinds of industrial, social and domestic facilities. Video surveillance systems are very important tools in security systems, and the requirements for the functionality of the segment are steadily growing.

Comprehensive security systems

The unified platform includes modules for security and fire equipment, access control and management, video surveillance or closed circuit television (CCT). Until recently, the functions of the latter were limited to video monitoring and recording of the situation at the facility and the surrounding area, archiving and storing data. Classic video systems have a number of significant disadvantages:

• Human factor. Ineffective operator performance when broadcasting a large volume of information.
• Impossibility of surgical intervention, untimely analysis.
• Significant time costs for searching and identifying an event.

The development of digital technologies has led to the creation of “smart” automated systems.

Strength is in the intellect

The basic principle of intellectual analytics is video analytics - a technology based on methods and algorithms for image recognition and automated data collection as a result of video stream analysis. Such equipment, without human intervention, is capable of detecting and tracking in real time specified targets (a car, a group of people), potentially dangerous situations (smoke, fire, unauthorized intervention in the operation of video cameras), programmed events and promptly issue an alarm signal. By filtering video data that is not of interest, the load on communication channels and the archive database is significantly reduced.

The most popular video analytics tool is a facial recognition system. Depending on the functions performed and the tasks assigned, certain requirements are imposed on the equipment.

Software and hardware

To ensure efficient operation of the system, several types of IP video cameras with different performance characteristics are used. The detection of an object in the controlled territory is recorded by panoramic cameras with a resolution of 1 megapixel and a focal length of 1 mm and scanning devices are pointed at it. These are more advanced cameras (from 2MP, from 2mm) that perform recognition by simple techniques(3-4 parameters). To identify an object, cameras with good image quality are used, sufficient for the use of complex algorithms (from 5 MP, 8-12 mm).

Most Popular software products for facial recognition "Face Intellect" (developed by House Control company), Face director (Sinesis company) and VOCORD FaceControl (VOCORD) demonstrate:

• High probability of object identification (up to 99%).
• Support for a wide range of camera rotation angles.
• Possibility of identifying faces even in dense pedestrian masses.
• Variability in the preparation of analytical reports.

Pattern Recognition Basics

Any biometric recognition systems are based on identifying the correspondence of the read physiological characteristics of an individual to a certain specified pattern.

Scanning occurs in real time. The IP camera broadcasts the video stream to the terminal, and the facial recognition system determines whether the image matches the photographs stored in the database. There are two main methods. The first is based on static principles: based on the results of processing biometric parameters, a electronic sample in the form of a unique number corresponding to a specific person. The second method models a “human” approach and is characterized by self-learning and robustness. Identification of a person from a video image takes into account age-related changes and other factors (presence of a headdress, beard or mustache, glasses). This technology allows you to work even with old photographs and, if necessary, with x-rays.

Face search algorithm

The most common technique for detecting faces is using Haar cascades (sets of masks).

The mask is a rectangular window with various combinations of white and black segments.

The mechanism of the program is as follows: a video frame is covered with a set of masks, and based on the results of convolution (counting pixels that fall into white and black sectors), the difference is calculated and compared with a certain threshold value.

To improve the performance of the classifier, positive (frames with human faces) and negative (without them) training samples are created. In the first case, the convolution result is above the threshold value, in the second - below. The face detector, with an acceptable error, determines the sum of the convolutions of all cascades and, if the threshold is exceeded, signals the presence of faces in the frame.

Recognition technologies

After detection and localization, the preliminary stage involves brightness and geometric alignment of the image. Further actions - feature calculation and identification - can be carried out using various methods.

When scanning a full-face face in a room with excellent lighting, good results are demonstrated by algorithms that work with two-dimensional images. By analyzing unique points and the distances between them, the face recognition system determines the fact of identification based on the difference coefficients between the “live” photo and the registered template.

3D technology is resistant to change luminous flux, permissible deviation from the frontal angle is up to 45 degrees. Here, not only points and lines are analyzed, but also the properties of surfaces (curvature, profile), and the metric of the distances between them. For such algorithms to work, maximum video recording quality with a frequency of up to 200 frames/s is required. The system is based on stereo video cameras with a matrix of 5 megapixels, high optical resolution and synchronization error reduced to a minimum. Additionally, they are connected by a special timing cable to transmit clock pulses.

State of the modern systems market

The former, due to their high cost, were developed only for government military facilities and only in the mid-90s became available to commercial organizations. Rapid development technologies and made it possible to increase the accuracy of systems and expand the scope of their application. The leading positions in the market of our country belong to American and Western European manufacturers of security systems. The top sellers are equipment from ZN Vision Technologies and Visionics corporations. The most promising among domestic developers are the research and products of Vocord, NTechLab, Soling, VisionLabs LLC and the TsRT group, which, among other things, are also engaged in adapting foreign complexes to Russian conditions.

Computer face control

The most extensive area of ​​application of contactless identification is the fight against terrorism and crime. The criminal's facial image is stored in a database. In places where there are large crowds of people (airports, train stations, shopping centers, sports institutions), the flow of people is being recorded in real time to identify wanted persons.

The next area is access control systems: a sample photograph on an electronic pass is compared with a model obtained as a result of processing data from video cameras. The procedure occurs instantly, without requiring any additional actions from those undergoing it (unlike a retinal scan or fingerprinting).

Another rapidly growing industry is marketing. An interactive billboard scans a person's face, determines his gender and age, and visualizes only those advertisements that will be potentially interesting to the client.

Trends and development prospects

Facial recognition systems are in great demand in the banking sector.

At the end of last year, the management of Pochta Bank, after installing 50,000 smart video cameras in its offices, managed to save millions of rubles by preventing fraud in the lending and payments segments. Experts say that by 2021 the necessary infrastructure network will be created and any operations at ATMs will become possible only after biometric identification of the client’s face.

In the next decade high technology will allow you to open a network of full self-service stores: the buyer walks in front of the display windows, selects the product he likes and leaves. The facial and image recognition system will determine the identity of the buyer, purchase and debit the required amount from his account.

Work is underway to create systems for recognizing psycho-emotional states. Analysis of human emotions will be in demand in multimedia fields: animation, cinema, and the computer game industry.

Facial recognition technologies are used in a wide variety of areas:

• ensuring safety in crowded places;
• security systems, avoiding illegal entry into the facility, searching for intruders;
• face control in the catering and entertainment segment, search for suspicious and potentially dangerous visitors;
• verification of bank cards;
• online payments;
• contextual advertising, digital marketing, Intelligent Signage and Digital Signage;
• photographic equipment;
• criminology;
• teleconferences;
• mobile applications;
• search for photos in large databases of photographs;
• tagging people in photos on social networks and many others.

IBM has released a database of 1 million photographs of faces for training biometric systems

2018

Facial recognition does not work in every second smartphone

In early January 2019, a Dutch non-profit organization tested 110 smartphone models and found that the facial recognition feature used to lock devices did not work properly on more than every second device.

A study carried out by Consumentenbond and its international partners found that 42 of the smartphones tested could be unlocked using just a photo of the phone's owner. Any photo will do, for example, obtained from social networks, from CCTV cameras or any other way.

Facial recognition software technology, available to owners of many smartphones running Android, has reached such a level of development that it no longer allows one to deceive oneself with a photo of the owner

The results of this study are of concern to users and security agencies. Using a printed photograph of the owner's face is the first test of the facial recognition feature that is used regular users and testers. But most importantly, this is the first trick that attackers will try to use to hack a smartphone protected by facial identification, before moving on to more complex attacks that include creating masks or 3D printed heads of the phone owner.

Any facial recognition system that fails the photo test is generally considered useless. According to Consumerbond, Asus, BlackBerry, Huawei, Lenovo, Nokia, Samsung, Sony and Xiaomi models failed these tests. In the case of Sony, absolutely all models failed the test. Another six models - Honor and six LG models - were tested only in the "strict" mode. While this test may lead users to conclude that it's not a good idea to enable facial recognition, 68 devices, including Apple's flagship iPhone XR and , survived this simple attack, as did many other high-end Android models from Samsung, Huawei, OnePlus and Honor.

A full list of models who passed the photo test can be found on the Consumerenbond website.

The most popular facial recognition systems in China

One of the most common facial recognition software is Face++, which is used for access control everywhere from Beijing train stations to an Alibaba office building.

Alibaba itself has developed own systems, which will be used in the Shanghai subway to identify passengers using their face and voice.

Police officers monitoring security at a Chinese railway station wear special sunglasses with facial recognition functionality. The device is capable of identifying a person in 100 milliseconds and has more than once helped law enforcement agencies in catching criminals.

In Shenzhen, China, a camera for recording violations by pedestrians was installed for the first time in the world. It is installed at one of the busy crossings in the city and monitors people crossing the road at a prohibiting traffic light. The camera uses facial recognition technology to determine the identity of the intruder.

College entrance exams across the country use facial and fingerprint recognition to ensure test takers are real students.

After a number of child abductions, some kindergartens are opening their doors only to people whose faces are registered in the system. In one of the kindergartens, more than 200 cameras were installed to ensure security.

Even some toilets have installed machines with facial recognition. The machine dispenses 60 cm of toilet paper to one person no more than once every nine minutes.

Alibaba has cashless Hema stores where users scan their face and enter a phone number to make payments through Alipay.

Alibaba together with a hotel manufacturer information systems Shiji has installed a facial recognition system for check-in at 50 hotels. Chinese tourists using the online travel agency Fliggy (owned by Alibaba) can first book a hotel there, then quickly check into the hotel using a “mask” of their face and make a deposit.

Beijing decided to fight illegal rental of public housing with the help of smart locks that recognize owners by face

At the end of December 2018, it became known that “smart” locks with facial recognition technology were being introduced at an accelerated pace in public housing in Beijing. With their help, local authorities are strengthening measures against the illegal re-renting of public housing provided to low-income families at preferential rates.

Smart lock with facial recognition

It is expected that by the end of June 2019, locks with a built-in face scanning system will be used in all programs for providing preferential state housing in Beijing with the participation of 120 thousand tenants, The South China Morning Post reports, citing the Beijing edition of The Beijing News.

By comparing the information obtained by scanning the faces of visitors with images from a stored database, the system recognizes the owners and does not open the doors to strangers, Shan Zhenyu, director of the information center at the Beijing State Housing Center, told Beijing News in an interview.

In addition, the system can be used to look after lonely elderly people. If the elderly person does not leave or enter the home for a certain period of time, a notification will be sent to the property manager to check on them.

In large cities like Beijing, housing rentals are very expensive. On average, a rented apartment in the Chinese capital costs about 5 thousand yuan per month (about $730), while rent for public housing can be less than 2 thousand yuan per month ($290).

Beijing authorities hope that smart locks that recognize owners by face will improve security, prevent illegal subletting and ensure that only people in genuine need benefit from benefits.

As of the end of 2018, smart locks with facial recognition are used in 47 public housing programs in Beijing. With their help, about 100 thousand scanned images of the faces of tenants and members of their families were obtained.

Chinese Airbnb installs smart locks with facial recognition in homes

Failure in London. The facial recognition system in the subway does not recognize anyone

At the end of December 2018, it became clear that the facial recognition system deployed on the London Underground would not recognize anyone. London police officers have been criticized for using unmarked vans to test controversial and inaccurate automatic facial recognition technology on Christmas shoppers. Read more.

Facial recognition toilets in China reduce toilet paper consumption

At the end of 2018, it became known about the growing number of public toilets in China with a facial recognition system that saves toilet paper.

In December, such a toilet opened in Baotu Spring Park in the city of Jinan (Shandong Province), located 400 km south of Beijing. In this restroom there is a machine that dispenses toilet paper after scanning your face. In one approach, the device dispenses approximately 70 cm of paper, and to receive an additional portion of a sanitary product, the same person needs to wait 9 minutes and again bring his head to the camera for identification.

To unlock a smartphone, hackers and police print the owner's head on a 3D printer

Facial recognition system launched at 14 American airports

On August 20, 2018, a facial recognition system was launched at 14 American airports. The U.S. Customs and Border Patrol (CBP) spoke about its effectiveness.

According to the department's website, on August 22, a 26-year-old passenger who flew into Washington Dulles Airport from Sao Paulo (Brazil) presented a French citizen's passport at the checkpoint. However, the biometric system revealed that the man's face did not match the photo in the document.

At Washington airport, a facial recognition system caught a man trying to enter the United States with someone else's passport.

When the arrival in the United States was sent for additional inspection, he was “obviously nervous” and, as it turned out, for good reason. In his shoe they found an identification card in the name of a citizen of the Republic of Congo, who in fact was the detainee. Now he faces imprisonment for trying to enter the United States with false documents.

British police facial recognition systems proved useless

In May 2018, it became known about big problems in the facial recognition systems used by British police. As a result, it may be filed large number claims - this issue has become a “priority” for the Information Commissioner’s Office, the BBC quotes the words of the regulator’s representative Elizabeth Denham.

British human rights organization Big Brother Watch has published research showing a “staggering” number of innocent people turned into potential criminals by facial recognition technology.

Thus, from May 2017 to March 2018, the system produced 2,685 matches of people with the suspect database for the South Wales Police, but 2,451 of them turned out to be false.

London law enforcement used facial identification technology at Notting Hill Carnival in 2017. The system's readings were incorrect in 98% of cases when a signal was triggered that a suspect from the police database had allegedly been spotted. The solution is designed in such a way that when a possible lawbreaker is identified, a signal is sent to the duty station at the nearest police station.

The police began to blame the cameras that produced low-quality images and the fact that the system was used for the first time, but the result did not improve in the subsequent 15 events (football matches, festivals, parades) during which the technology was used. Only on three of them did the system not make a single mistake.

The police also said that during the nine months of operation of the facial recognition system, it correctly identified more than 2 thousand people, which led to 450 arrests. However, no one was wrongly imprisoned. This is explained by the fact that in addition to the work of the algorithms, people are involved in the work, who check the responses and make the final decisions.

Scientists have invented a new way to deceive facial recognition systems

Every day, facial recognition systems are becoming more complex and are increasingly used in everyday life; for example, last year Apple released the iPhone X smartphone equipped with the Face ID biometric system. However, such systems can be deceived, in particular, with the help of infrared LEDs. Infrared rays are not visible to the naked eye, but most cameras can detect infrared signals.

Chinese researchers have created a baseball cap equipped with miniature infrared LEDs, which are placed in such a way that the infrared rays falling on the wearer's face help not only to hide his identity, but also to "impersonate another person for facial recognition-based authentication." . This task is more complex and requires the use of a deep neural network to recognize a static image of a face and correctly project infrared rays onto the impostor’s face.

To test their theory, the researchers used photographs of four random people, and they were able to fool facial recognition systems 70% of the time, provided there was little resemblance between the victim and the impostor.

“Based on our findings and attacks, we can conclude that current facial recognition technologies are difficult to call secure and reliable in critical scenarios such as authentication and surveillance,” the researchers concluded. They also added that infrared LEDs could be hidden not only in baseball caps, but also in umbrellas, hair or wigs.

Russian twins demand 20 million from Apple because iPhone X does not see the difference between them

Twin brothers from Vladimir - 26-year-old Alexander and Ilya Tunchik - sent a complaint to the Russian office of Apple due to the fact that the Face ID facial recognition system on their iPhone X smartphones identically identifies both young people, thereby, in their opinion, violating the protection of personal data.

Offended users demand that the company improve the technology, as well as compensate moral damages in the amount of 20 million rubles, Roman Ardykutsa, a lawyer representing the brothers’ interests, told TASS in January 2018.

“The twins purchased... an iPhone X precisely in order to use the screen unlocking feature using their faces. To their disappointment, each device recognizes both brothers, which they were not warned about when purchasing; this information is not in the instructions. That is why the applicants are asking the company to refine the technology,” he explained.

2017

Facial recognition in retail

In November 2017, CNBC released a story about the introduction of facial recognition systems in stores. Retailers use such technologies to collect customer data and tailor offers based on relevant data.

In retail, facial recognition is used mainly to motivate customers. For example, if a person is recognized at the entrance to a store and their purchase history is seen, then the store employees know better what to offer him. So, if he bought a TV at an electronics store, the employee will recognize him, call him by name and offer to buy a new remote control.

According to Hong Kong IT company Jardine One Solution (JOS), many retailers are using facial recognition capabilities to collect data about visitors to their stores.

JOS itself helps retailers with facial recognition to build customer profiles and track their actions at the point of sale. We are talking about data such as the number of visitors, their age, gender, ethnicity. Such information helps stores better understand the flow of customers and select personalized offers for them, Lunt noted.

For example, using the analysis of data coming from facial recognition systems, you can select the music playing in the trading floor.

JOS says that all customer data received is anonymous, but the issue of confidentiality remains relevant. Technology is not preventing the adoption of such systems, but there are concerns related to personal data and culture, admits Mark Lunt.

He added that retailers spend a lot of money on preventing data leaks and protecting information. The scandal involving the theft of data from millions of Uber customers shows that companies cannot feel safe and users must be careful when disclosing personal information, says the managing director of JOS.

Founder and general manager HeadCount, which offers traffic monitoring and improvement services to stores, Mark Ryski says biometric data, including that generated by facial recognition systems, is sensitive and has great potential - especially for security and safety purposes. improving the quality of customer service.

An example of using a facial recognition system in stores

According to Brennan Wilkie, senior vice president of customer strategy at InMoment, there is indeed a lot of potential for using facial recognition equipment in retail environments. For example, such devices are able to compare a customer's facial expression in a store with data about him, his brand loyalty and other purchases. To mitigate user privacy concerns, stores need to demonstrate to customers what benefits they receive, just as they did with self-checkout counters or chip-enabled bank cards, he said.

According to the forecast of the analytical company MarketsandMarkets, the global market for facial recognition systems will reach $6.8 billion by 2021.

Authorization in iPhone X by face was hacked using a mask for $150. Video

How to bypass the face scanner on Samsung Galaxy Note 8

Web designer Mel Tahon posted a video on Twitter showing how to easily bypass the face scanner on Galaxy Note 8. In his experiment, Tahon holds two Note 8s facing each other, one with his photo on it, and the other with the face scanning system turned on.

Samsung Galaxy S8 biometric security trick

Researchers were able to pass off a white man as Milla Jovovich in almost 90 percent of cases. An Asian woman wearing special glasses was mistaken by the computer for a man from the Middle East in the same percentage of cases.

They also tested their method on the commercial software Face++, which is used by Alibaba to authorize payments. In this case, they did not sit the person wearing glasses in front of the camera, but first took a photo of him wearing glasses and then loaded it into the program. As a result, they were able to pass off one person as another in 100 percent of cases.

US public organizations against facial recognition

A coalition of 52 civil society and human rights organizations sent a letter to the Department of Justice asking it to investigate the excessive use of facial recognition technology in law enforcement. The coalition is also concerned about the uneven accuracy of machine recognition of faces of different racial backgrounds, which could become the basis for racism on the part of law enforcement officers.

These technologies are especially abused by local police, state police and the FBI, the letter says. The coalition is asking the Justice Department to prioritize investigations into police departments that are already under investigation for bias against people of color.

The basis for the request was the results of a study by the Center for Privacy and Technology at Georgetown University School of Law. The study found that the faces of half of the US adult population were scanned by government identification software under various circumstances.

The researchers note that in the United States today there are no serious regulations governing the use of this software. According to Alvaro Bedoya, director of the Center and co-author of the study, once a person is photographed with a driver's license, they are already included in the police or FBI database. This is especially significant given that facial recognition can be inaccurate and can harm innocent citizens.

Examples of projects in HSBC, MasterCard and Facebook

The service will be available for corporate clients NSBC. Via banking mobile application they will be able to open accounts with one click of a selfie. The bank confirms the client’s identity using a facial recognition program. The photo is compared with images previously uploaded to the system, for example, from a passport or driver’s license. It is assumed that new service eliminates the need to remember digital codes and will reduce identification time.

To use this option, users will need to download special application to your computer, tablet or smartphone. Then look into the camera or use the device's fingerprint scanner (if the device has one). However (at least for now), users will still need to additionally provide data from their bank card. Only if additional identification is required will users be able to use the above option.

With this new approach, MasterCard intends to protect users from fraudulent online transactions that are carried out using stolen user passwords, as well as provide users with a more convenient authorization system. The company reported that 92% of people who tested this new system preferred it to traditional passwords.

Some experts question the security of information to prevent cybercriminals from easily obtaining a user's fingerprints or photo of their face if a transaction occurs while using a public Wi-Fi network insecurely.

Cybersecurity experts say the system should include multiple layers of security to prevent the potential theft of users' facial photos. After all, online payments are an attractive target for cyber criminals.

At the end of 2015, a group of experts from the Technical University of Berlin demonstrated the ability to extract the PIN code of any smartphone using a user's selfie. To do this, they read this code, which was displayed in the user's eyes when he entered it on his OPPO N1 phone. It is enough for a hacker to simply seize control of front camera smartphone to perform this rudimentary attack. Could a cybercriminal take control of a user's device, take a selfie, and then make online payments using the typed password that the hacker saw in the eyes of his victim?

MasterCard insists its security mechanisms will be able to detect such behavior. For example, users would need to flash for an app to show a “live” image of a person, rather than a photo or pre-shot video of them. The system matches the user's facial image, converting it into a code and transmitting it via a secure protocol over the Internet to MasterCard. The company promises that this information will be securely stored on its servers, and the company itself will not be able to reconstruct the user's face.

In the summer of 2016 it became known that Researchers bypassed the biometric authentication system using a photo from Facebook. The attack was made possible due to potential vulnerabilities inherent in social resources.

A team of researchers from North Carolina State University have demonstrated a method for bypassing security systems built on facial recognition technology using accessible photos of social network users. As explained in the specialists' report, the attack was made possible due to potential vulnerabilities inherent in social resources.

“It is not surprising that personal photos posted on social media can pose a privacy risk. Most major social networks recommend that users set privacy settings when posting photos on the site, but many of these photos are often available to the general public or can only be viewed by friends. In addition, users cannot independently control the availability of their photos posted by other subscribers,” the scientists note.

As part of the experiment, the researchers selected photographs of 20 volunteers ( Facebook users, Google+, LinkedIn and other social resources). They then used these images to create 3D models faces, “revitalized” them using a number of animation effects, applied skin texture to the model and adjusted the gaze (if necessary). The researchers tested the resulting models on five security systems, four of which were fooled in 55-85% of cases.

According to the company report Technavo(winter 216) one of the key trends that has a positive impact on the market for biometric facial identification technologies ( facial recognition), is the introduction of multimodal biometric systems in sectors such as healthcare, banking, financial sector, securities and insurance sector, transportation sector, road transport, as well as in the public sector.

The founder of the project, Benjamin Levy, said that thanks to high level security IsItYou will be able to recognize 99999 out of 100 thousand cases of deception. Levy tried to convince banks to implement his system as early as next year. It will be used to conduct financial transactions.

Google already uses facial recognition in Android. This way you can unlock a device running this mobile OS. However, developers have repeatedly argued that facial recognition is not secure enough compared to classical methods. In this regard, experts doubted the statements of Benjamin Levy.

Marios Savvedes from Carnegie Mellon University is researching facial recognition. He believes that IsItYou's self-conducted security test cannot be reliable.

World expert in the field of biometrics, Dr. Massimo Tistarelli, shares the same opinion. He said that a full-scale research project, Tabula Rasa, is underway in Europe, the main goal of which is to develop anti-fraud protection for biometric identification methods. According to him, before entering the market, a number of independent studies should be conducted to confirm the effectiveness of the product.

Column

It threatens human security and civil rights, so partial regulation should be replaced by a complete ban. While the world is fascinated by the benefits of facial recognition technology, some security experts believe that it holds a great evil for humanity. Law and computer science professor Woodrow Hartzog and philosophy professor Evan Selinger offered their views on technology control methods in a Medium article.

The people of Troy would be delighted

It is very easy to be swayed by seemingly tempting, but in fact erroneous, ideas about what the future of humanity will be like in a world that has unlocked the full hidden potential of facial recognition technology. People will be able to instantly receive information about strangers, they will no longer have to remember a ton of passwords or be afraid of forgetting their wallet. It will be possible to easily find events from a certain person in archives of photographs and videos, quickly search for missing people or criminals, and make public places safe.

It would seem that technology brings only advantages, absolute justice will reign in the world, and the most incredible ideas of humanity will be realized. But none of the surveillance mechanisms invented by mankind is as dangerous as facial recognition technology.

Seduced by this utopian vision, people will bring facial recognition technology into their homes and devices, allowing it to take center stage in ever new aspects of life. This will mean that the trap has slammed shut, and then the unpleasant realization will come that the technology was a kind of Trojan horse. This ideal tool of oppression is too good not to be used by governments to establish authoritarian controls and all-encompassing regimes that will destroy the concept of privacy.

This Trojan horse must not enter the city.

Current discussions

The American Civil Liberties Union, along with 70 other human rights organizations, demanded that Amazon stop providing facial recognition technology to the government, and also called on Congress to impose a moratorium on its use by the government. The media also joined in and expressed their concern. For example, the editorial board of the Washington Post believes that Congress is obliged to immediately intervene in the situation. MPs also have good reason to wonder: some of them are linked to Amazon's facial recognition software by criminals.

The editors of The Guardian were not left out either. Microsoft President Brad Smith wrote in a blog post asking the US government to regulate facial recognition technology:

"The only one reliable way to control the use of technology by the government means that it independently and taking into account possible circumstances controls its use. We believe there is an urgent need today for a government initiative to monitor the lawful use of facial recognition technology, based on the decision of a bipartisan panel of experts."

The opinions of company leaders are important, as are legislation that restricts the use of technology. But only partial support and carefully written instructions will never be enough. Laws could be of great benefit, but they will most likely begin to be introduced when the technology becomes much cheaper and easier to use. Smith points out that Microsoft called for a national law in this area back in 2005. More than ten years have passed, but Congress has not passed such a law.

If facial recognition technology continues to be developed and implemented in life, a gigantic infrastructure will emerge that will engulf humanity. History has shown that widespread attention to success, fear of insecurity, and an intoxicating sense of power can lead to deception, shifting corporate values, and ultimately systematic abuse of technology.

The future well-being of humanity is only possible if facial recognition technology is banned before it becomes too entrenched in human life.

Why is there a ban?

The need to completely ban facial recognition systems is extreme. But some talented scientists, like Judith Donat, believe this position is incorrect. They propose a more technologically neutral tactic: banning specific actions, as well as identifying values ​​and rights that need to be protected. This approach makes perfect sense for almost all digital technologies.

But none of the surveillance mechanisms invented by mankind is as dangerous as facial recognition technology. It is the missing piece of an already dangerous human surveillance infrastructure, developed because governments and private businesses need this infrastructure. And if technologies become dangerous to such an extent, and the ratio of benefit and harm becomes so distorted, it’s time to think about categorical prohibitions. Some types of dangerous digital technologies, such as spyware, are already prohibited at the legislative level. Facial recognition technology carries much greater risks, and it would do well to receive special legal attention. What is needed is a specific ban on a robust, holistic, values-based and largely technology-neutral basis. regulatory framework. Such a system will help avoid regulatory situations where legislators try to catch up with technological trends.

Surveillance using facial recognition systems is inherently oppressive. The existence of such systems, which themselves are often hidden from human sight, is a violation of civil liberties because people behave differently if they suspect that they are being watched. Even laws that guarantee strict protective measures will not prevent the oppressive feeling that a person's opportunities for self-expression will be undermined.

Here are examples of the misuse and destructive effects of facial recognition technology:

• disproportionate attention to people of color, other minorities and vulnerable peoples;
• replacing the presumption of innocence with the principle of “people whose guilt has not yet been proven”;
• the spread of violence and cruelty;
• denial of fundamental rights and opportunities, such as protection from arbitrary government tracking of a person's movements, habits, relationships, interests and thoughts;
• the continuous “work” of the law - as a permanent preventive measure;
• destroying the concept of storing information “practically obscure” when the data is in open access, but are stored in various sources and are extremely difficult to find;
• the spread of "surveillance capitalism".

As facial recognition researcher Claire Garvey notes, errors in it can have fatal consequences:

“What happens if a system like this fails? If CCTV systems go wrong, an innocent person will be chased, interrogated, or may even be arrested and charged with a crime. Or portable cameras with a facial recognition system for police officers: if the system points to a person who allegedly could pose a danger to society, the police officer will have to instantly decide whether to use a weapon. Innocent people may be harmed as a result of a false alert.”

Among others, there are two reports that address many of these issues in detail: a very valuable paper on the use of facial recognition by law enforcement, published by Electronic Frontier Foundation senior attorney Jennifer Lynch, and a study by the Center on Privacy & Technology at Georgetown University.

Despite the problems described in the reports, not everyone is convinced that a ban is truly necessary. After all, other technologies pose no less a threat: geolocation data, information from profiles on social networks, results search queries and many other sources of information about users can be used to create a detailed portrait of them. But facial recognition still carries a different danger and stands apart even in comparison with biometric data: fingerprints, DNA samples or retinal scans.

Systems that process facial images have five distinctive features that give every reason to ban them. Firstly, the face is difficult to hide or change. Faces cannot be encrypted like data on digital media or in emails or text messages. They can be filmed using remote cameras, and the cost of the technology itself and storing images in the cloud is constantly decreasing, which leads to the increasingly widespread use of such monitoring systems.

Secondly, there are databases of names and faces, for example for driver's licenses, or social media accounts, which can be accessed very easily.

Third, unlike typical surveillance systems, which often require expensive equipment or new data sources, the input data for facial recognition is ubiquitous and comes directly from the moment cameras are captured.

Fourth, a turning point. Any database of individuals to identify individuals arrested or caught on camera can be “matched” with any other database in real time using a few lines of code by connecting to portable police cameras or CCTV systems. New York State Governor Andrew Cuomo accurately noted the reasons for the proliferation of facial recognition technology, arguing that simply scanning car license plates will seem trivial compared to the capabilities of cameras with built-in technology: “The system reads the license plate to identify the violator, but the fines are far from the biggest benefit of this equipment. We are moving to facial recognition technology, and now the system will be able to scan the driver’s face and check it against databases, which opens up completely new perspectives.”

Fifth, the face, unlike fingerprints, gait, or retinal images, is a central element of a person's identity. A person is an intermediary between virtual and real life person, link between actions that a person performs anonymously, under his own or someone else’s name. It may easily seem that there is no need to protect facial privacy like any other private information, because in life people usually do not cover their faces. Except in countries where women are required to wear the burqa, people with a hidden face are viewed with suspicion.

Ensuring the privacy of a person's face is indeed necessary because in the past, people developed institutions and values ​​related to the protection of private information during periods when identifying strangers was generally quite difficult. Due to biological characteristics, human memory is limited, and without a technological add-on, he can remember only a small number of faces. And given the size and distribution of the population, a person will not meet many new people in his life. These restrictions created a kind of blind spot, giving people a good chance of getting lost in the crowd.

Recent U.S. Supreme Court decisions regarding the Fourth Amendment (which prohibits unreasonable searches and seizures and requires the issuance of search warrants by a court on a showing of probable cause) indicate that the fight to protect privacy in public places remains relevant. This summer, in one of the trials, the court decided that geolocation data from mobile phones are subject to the Constitution, and information that a person wishes to keep secret, even if it is publicly available, may be protected by the Constitution.

Why facial recognition technology is not subject to legal regulation

Due to the fact that facial recognition technology poses a huge threat, society cannot leave its regulation to chance. Potential profitability will stimulate the emergence of ideas to realize the maximum potential of the technology, and individual companies will advance their interests in this direction.

Society also cannot wait for the rise of populists. Facial recognition technology will continue to be “sold” as part of the newest and most advanced applications and devices. Apple is already calling Face ID the best feature of the latest iPhone. The same goes for ideologically charged news reports in which facial recognition technology is hailed as the solution to all problems.

Finally, society should not rely unduly on traditional regulatory methods. The peculiarities of facial recognition technology make it difficult to contain it within the framework of measures that define legal and illegal uses and try to accommodate its potential usefulness to society and deterrence factor for attackers. This is one of the few examples where it is necessary to introduce a complete ban.

At the moment, there are very few projects to control facial recognition technology and even fewer to limit it. There are decent biometric data laws in Illinois and Texas, but they follow a conventional regulatory strategy that requires entities collecting and using this data to comply with a set of basic information practices and privacy protocols. These include the requirement to obtain informed consent for the collection of biometric data, mandatory protection and retention limits, prohibitions on its use for profit, restrictions on transfer rights to third parties, and private causes of action in the event of violation of these standards.

Proposed facial recognition laws are similar. The US Federal Trade Commission recommends the same mechanism for technology: warning people about its use, giving them a choice, and fairly limiting the use of their data. The Electronic Frontier Foundation's report, which focuses on enforcing these laws, makes similar, albeit more profound, suggestions. For example, create clear rules for the use, distribution and security of data; introduce restrictions on data collection and storage; a ban on including several types of biometric data in one database; mandatory notification, inspections and independent supervision. In its draft facial recognition legislation, the Center on Privacy & Technology at Georgetown University proposes to significantly limit government access to facial databases, as well as the use of real-time facial recognition technology.

Unfortunately, most current and proposed requirements are procedural in nature. And ultimately, they will not stop the spread of the technology itself and the development of the corresponding infrastructure. The first thing to note is the falsity of some of the underlying assumptions about consent, notice, and choice that exist in existing laws. Informed consent as a mechanism for regulating surveillance and data processing is completely useless. Even if people had complete control over their data, they would still not be able to take full advantage of it.

Yet lawmakers and the industry itself are trying to move the needle. But these regulations, like most privacy regulations in the digital era, have many gaps. Some laws only concern the collection or storage of data and do not address how it is used. Others apply only to companies or the government and are so ambiguous that they avoid consequences for various illegal actions. And realizing the benefits of much-touted facial recognition technology will require more cameras, better infrastructure and vast databases.

The Future of Facial Recognition Technology

Facial recognition technology opens up endless possibilities for tracking information about a person’s identity and movements. And also almost instantly save, distribute and analyze it. The development of this technology in the future may lead to the fact that the confidentiality of a person's private information will be constantly violated. The well-being of humanity is only possible if a ban is introduced on facial recognition technologies before these systems become too entrenched in everyday life. Otherwise, people will only be familiar with a world in which, every time they appear in a public place, they will be automatically identified, entered into their profile information and, possibly, used. In such a world, those who oppose facial recognition technology will be discredited, silenced, or eliminated.

The anniversary iPhone X received one of the most extraordinary features among its competitors. The flagship can recognize the owner's face, and instead of Touch ID and the Home button, engineers integrated the TrueDepth camera and the Face ID function.

Fast, instant and without the need to enter passwords. This is how you can unlock your iPhone X today.

Apple is known for always looking to the technological future long before the next feature becomes standard. In the case of the iPhone X and the face scanner, the company is confident that facial recognition is the future.

Let's figure out whether Apple is mistaken or our faces - this is a sure pass to the digital future.

😎 The Technology section is published every week with the support of re:Store.

So how does facial recognition work?

Facial recognition technology requires several components to work. Firstly, the server itself, on which both the database and the prepared comparison algorithm will be stored.

Secondly, a well-thought-out and trained neural network, which was fed millions of photographs with marks. Such networks are easy to train. They upload the photo and present it to the system: “This is Viktor Ivanov,” then the next one.

The neural network independently distributes feature vectors and finds geometric patterns of the face in such a way that it can then independently recognize Victor from thousands of other photographs.

The same FaceN technology, which we will talk about below, uses about 80 different numerical characteristics.

Why are people suddenly talking about facial recognition?

In mid-2016, the Internet literally exploded with an application of the same name. Using neural networks, the developers managed to make the wildest dream of social network users come true.

When you saw a person on the street, you could take a photo of him on your smartphone, send the photo to FindFace, and in a few seconds find his page on VKontakte. The algorithm was improved, updated and recognized faces better and better.

It all started with recognizing dog breeds from photographs. The author of the FaceN recognition technology and the Magic Dog application is Artem Kukharenko. The guy quickly realized that this technology was the future and began development.

After success FindFace applications, founder of the development company N-Tech.Lab Kukharenko was once again convinced that facial recognition is interesting in almost any industry:

In May 2016, N-Tech.Lab began testing the service together with the Moscow government. Tens of thousands of cameras were placed throughout the capital, which identified passers-by in real time.

Trustory. You simply walk through the yard in which such a camera is installed. A database of criminals and missing people is connected to it. If the algorithm determines that you are similar to the suspect, the police officer immediately receives a warning.

Of course, a person can be immediately found in social network and hit any bases. Now imagine that such cameras are installed along the perimeter of the entire city. The attacker will not be able to escape. There are cameras everywhere: in courtyards, at entrances, on highways.

How are things going with facial recognition in Russia?

You will be surprised, but since mid-2016, Moscow mayors have been actively implementing a facial recognition system throughout the city.

To date, more than 100 thousand cameras capable of recognizing faces have been installed at the entrances of Moscow high-rise buildings alone. More than 25 thousand are installed in yards. Of course, the exact numbers are classified, but rest assured, active control is spreading faster than you might imagine.

In the capital, facial recognition systems are installed everywhere: from squares and crowded places to public transport. Since the installation of the systems, more than ten criminals have been detained, but this is only according to official data.

All cameras constantly exchange information with the One computing center Department of Information Technology. Suspicious alerts are immediately checked by law enforcement agencies.

And this is just the beginning. At the end of last year, a similar control system began to be tested on the streets of St. Petersburg. The convenience of the technology proposed by FindN is that it is not necessary to install any special cameras.

The image from standard CCTV cameras is processed by a “smart” algorithm and the real magic happens there. According to current data, the recognition accuracy of FindFace today varies between 73% - 75%. The developers are confident that they will be able to achieve 100% results in the near future.

How did facial recognition come about?

Initially, any type of biometric identification was used exclusively internally law enforcement agencies and services where safety is a priority. In just a few years, measuring anatomical and physiological characteristics for personal identification has become standard in almost all consumer gadgets.

There are many types of biometric authentication:

And it is the latter technology that is especially interesting, since it has several advantages over others.

The prototype of facial recognition technology in the 19th century was first “portraits by description”, and later – photographs. This way the police could identify the criminals. In 1965, a semi-automatic facial recognition system was developed specifically for the US government. In 1971, the technology was returned to, identifying the basic markers necessary for facial recognition, but not for long.

Since then, intelligence agencies have still preferred proven fingerprinting technology as their main biometric identifier.

And all because technology did not allow any interaction with human facial features. Ultra-precise lasers, infrared sensors and powerful processors, as well as the recognition systems themselves, did not exist at that time.

With the advent of powerful computers, almost all departments are returning to identification through facial scanning. The technology boomed in departments and special agencies in the mid-2000s, and last year the technology began to be used for the first time in consumer devices.

Where is facial recognition technology used today?

In smartphones

The popularization of facial recognition technology began with Apple's flagship. The iPhone X set the trend for the coming years and OEMs have actively begun integrating Face ID analogues into their devices.

In banks

Biometric facial recognition has been used in the United States for several years. Now the technology has reached Russia. In 2017 alone, thanks to the implementation of this system, it was possible to prevent more than 10 thousand fraudulent transactions and save an amount of 1.5 billion rubles.

Facial recognition is used to identify the client and make a decision on the possibility of issuing a loan.

In stores

The retail segment uses technology in its own way. So, if you bought any household appliances in the store, and after some time you return to it for another purchase, the facial recognition system immediately identifies you at the entrance. The seller will immediately receive information from the database and find out not only your name, but also your purchase history. The seller's further behavior is easy to predict.

In the life of cities

This is exactly what technology is developed and developed for. From stadiums to movie theaters, wherever there are huge numbers of people, identification is especially important. Today, facial recognition technology makes it possible to prevent riots and terrorist attacks.

Which companies are interested in facial recognition?

Google, Facebook, Apple and other IT giants are now actively purchasing projects from developers involved in facial recognition. They all see great potential in the technology.

This is only a part of the officially announced deals. In fact, there are many more of them. In addition to integrating Face ID and analogue technologies into smartphones, leading IT companies have much greater plans for the use of facial recognition.

What a future with facial recognition will look like

With the benefits of face scanning technology in smartphones and electronic devices, we’ve already figured it out, then let’s look into the near future and imagine one day in the life of a person who finds himself in a city where facial recognition cameras are installed everywhere.

Good morning! Smile, the smart home system is looking at you. Hmm, master, I drank a lot yesterday - I can see it in my face, I had difficulty recognizing it. So, next to my wife, Barsik is finishing his evening food in the hallway. There are no strangers. Amazing.

One glance at the coffee maker at a distance “a little closer than usual” and your medium-strength Americano with slightly warm milk is ready. Oops, someone is at the door! Oh, this is my favorite mother-in-law. Come in, the door is open for you - not a single recognition system in the world will forget your face.

You get ready and approach the elevator. No, no, this recognition system already knows that you prefer to sit in the outer elevator, so it has already been called.

Seeing you from afar, the 500-horsepower electric car automatically adjusted the reach of the steering wheel and adjusted the position of the seat. The door is open - take a seat.

While manufacturers of autopilot systems are unsuccessfully trying to convince legislation of the need to introduce unmanned vehicles, try not to violate traffic rules. Surveillance cameras are everywhere, and paying a fine is inevitable. After all, you are the one driving, and as soon as you press the accelerator pedal to the floor, a speeding fine will be debited from your bank card.

Finally, we are at the office building of the very company that is introducing facial recognition technology into the infrastructure of Russian cities. Yes, that's your job. The control is tight, but you don’t have to worry - while you were parking the car, the cameras already recognized you.

Work has become more difficult: along the entire perimeter of the office there are recognition cameras that “see” who is doing what, and at the same time they can read emotions. In short, fooling around in the workplace will not work.

There is perhaps no other technology today that is surrounded by so many myths, lies and incompetence. Journalists who talk about technology lie, politicians who talk about successful implementation lie, most technology sellers lie. Every month I see the consequences of people trying to implement facial recognition into systems that can't handle it.

The topic of this article has been sore for a long time, but I was still too lazy to write it. A lot of text that I have already repeated twenty times to different people. But, after reading yet another pack of trash, I decided it was time. I will provide a link to this article.

So. In this article I will answer a few simple questions:

Where do you think the creators of the algorithms got these bases from?

A little hint. The first NTech product they have now is Find Face, a search for people on VKontakte. I think no explanation is needed. Of course, VKontakte is fighting bots that pump out everything open profiles. But, as far as I heard, people are still downloading. And classmates. And Instagram.

It seems like with Facebook - everything is more complicated there. But I’m almost sure that they also came up with something.
So yes, if your profile is public, then you can be proud, it was used to train algorithms;)

About solutions and about companies

Now the remaining leaders are 1-2 Chinese companies and 1 American, Vocord has lost something in the ratings.

Other Russian ones in the rating are itmo, 3divi, and intellivision. Synesis- Belarusian company, although some of them were once in Moscow, about 3 years ago they had a blog on Habré. I know about several other solutions that they belong to foreign companies, but the development offices are also in Russia. There are also several Russian companies that are not included in the competition, but which seem to have good solutions. For example, the MDGs have. Obviously, Odnoklassniki and Vkontakte also have their own good ones, but they are for internal use.

In short, yes, it’s mostly us and the Chinese who have the same facial expressions.

NTech was the first in the world to show good parameters new level. Somewhere at the end of 2015. VisionLabs has only just caught up with NTech. In 2015 they were the market leaders. But their decision was of the last generation, and they began to try to catch up with NTech only at the end of 2016.

To be honest, I don't like either of these companies. Very aggressive marketing. I have seen people who were sold a clearly inappropriate solution that did not solve their problems.

From this side I liked Vocord much more. I once consulted some guys to whom Vocord very honestly said, “Your project will not work with such cameras and installation points.” NTech and VisionLabs happily tried to sell. But Vocord has disappeared recently.

Conclusions

Today, you cannot believe all the fairy tales that you can catch all the criminals, or observe everyone in the city. But it's important to remember that such things can help catch criminals. For example, to stop not everyone on the subway, but only those whom the system considers similar. Place cameras so that faces are better recognized and create the appropriate infrastructure for this. Although, for example, I am against this. Because the cost of a mistake if you are recognized as someone else may be too high.