Systems of face recognition Facial recognition
Face recognition is an automatic localization of a human face on the image or video and, if necessary, identification of the identity of the person on the basis of the available databases. Interest in these systems is very big in connection with a wide range of problems which they solve. You look also DeepFake (from deep learning – "deep learning" and fake – "counterfeit") – a method of synthesis of the human image on the basis of artificial intelligence. It is used for consolidation and imposing of the existing images on video.
Sensing technologies of persons are applied in the most various spheres:
- security in places of big accumulation of people;
- the systems of protection, avoidance of illegal penetration on the territory of an object, search of malefactors;
- face control in a segment of a public catering and entertainments, search of suspicious and potentially dangerous visitors;
- verification of bank cards;
- online payments;
- contextual advertizing, digital marketing, Intelligent Signage and Digital Signage;
- photography equipment;
- mobile applications;
- search of a photo in big bases of pictures;
- mark of people in a photo on social networks and many others.
Apple is going to use the system of face recognition as phone unblocking – the selfie filmed by the owner of phone with the front camera will be compared to in advance loaded photo standard.
The main lack of a sensing technology of persons is quality degradation of recognition at
- deterioration in illumination;
- change of provision of the head or foreshortening.
There are several approaches for creation of an algorithm of face recognition.
Empirical approach was used at the very beginning of development of computer vision. It is based on some rules which are used by the person for detecting of the person. For example, a forehead it is normal more brightly, than the central part of the face which, in turn, is homogeneous on brightness and color. One more important sign is existence of parts of the face on the image – a nose, a mouth, eyes. Considerable reduction of section of the image where presence of the person is supposed is made for determination of persons, or perpendicular histograms are under construction. It is easy to implement these methods, but they are almost unsuitable with a large number of foreign objects against the background of, several persons in a frame or at change of a foreshortening.
The following approach uses the invariant signs characteristic of the image of the person. In its basis, as well as in the previous method, the attempt of the think system as the person lies the empiricist, i.e. The method reveals characteristic parts of the face, its border, change of a form, contrast, etc., integrates all these signs and verifies. This method can be used even at turn of the head, but in the presence of other persons or a heterogeneous background recognition becomes impossible.
The following algorithm is a detecting of persons using templates which are set by the developer. The person is represented a certain template or the standard, and the algorithm purpose – to make check of each segment on existence of this template, and check can be made for different foreshortenings and scales. Such system requires a set of labor-consuming calculations.
All modern sensing technologies of persons use systems studying using test images. For training bases with the images which are containing persons, and not supporting the person are used. Each fragment of the studied image is characterized as a vector of signs by which qualifiers (algorithms for object definition in a frame) define, this part of the image is a person or not.
As the system of face recognition in the subway of Moscow works
Moscow will install the system of face recognition of participants of protest actions. It will become the largest after China
At the beginning of September, 2019 it became known of the contract for the amount of 260 million rubles which the Moscow Department of Information Technology (DIT) signed with Sitronics for technical maintenance of mass actions. It is going to create the system of face recognition of participants of meetings. Read more here.
American Airlines began to use the system of face recognition instead of boarding passes
At the end of August, 2019 American Airlines began to use a sensing technology of persons that passengers could get into the airplane without boarding pass. The airline states that this system is absolutely safe and accelerates landing. Read more here.
The system of face recognition defined that the murderer wants to take the credit for a corpse of the victim
At the end of August, 2019 the Chinese police delayed the man accused of murder of the girlfriend when assumed software for face recognition that he tried to scan the person of the victim for the request for the credit.
The police consider that the 29-year-old man by the name of Zhang strangled the girl a rope on April 11, 2019 after the quarrel. It is supposed that then the murderer went to races, having hidden a body in a luggage carrier of the leased car. Besides, the murderer on behalf of the victim contacted her employers through an account of WeChat and received a holiday. Having arrived in the hometown of Sanming next day, he tried to submit the application for the credit, using the application under the name Money Station.
This system uses AI for check of the identity of applicants and asks them to wink to help verification process and identification. However the victim, certainly, could not wink. The suspicious request was right there sent for consideration to the operator. The staff of credit institution immediately contacted police when at repeated manual check detected bruises on a woman's face and the deep red line on her neck. One more catch was what software for recognition of a voice defined that the man, but not the woman addressed for receiving the credit.
On the basis of data retrieveds the prosecutor's office of China approved detention of the alleged murderer. It was caught on a remote farm where he tried to burn a body of the victim. Now he is accused of murder, counterfeit of the personality and theft of means of the victim. Though date of the final hearing of the case is not appointed yet, its parts shook the public. Many are recognized, as did not suspect as software on face recognition and voices can be used in such situations for the help to police officers.
At football stadium installed the system of face recognition of Panasonic which does not pass fans from the black list
At the end of August, 2019 the Japanese producer of electronics Panasonic announced that the surveillance cameras of the company mounted at football stadium of the Danish club Brøndby IF will include the software for face recognition now. New cameras will identify persons from the black list which prohibits access to the stadium territory. Read more here.
Britain starts the system of face recognition of the gone people
In the middle of July, 2019 it became known of start in Britain of the system of face recognition of the gone people. For testing of technology the Ministry of Internal Affairs addressed police in the Counties of West Midlands, Kent and also the British traffic police.
According to the The Telegraph edition with reference to National fight against crime agency, annually vanishes in Britain about 180 thousand people, and every 200th of them is a child.
Law enforcement agencies will use thousands of surveillance cameras for identification of people who can be wanted. Their persons will be compared with the corresponding database.
The authorities decided to expand face recognition systems capabilities, despite big criticism concerning the accuracy of work of technology.
The research published in June, 2019 by the University of Essex by request of Scotland Yard showed that 81% of the "suspects" marked using a sensing technology of persons are innocent and that the vast majority of the "identified" people actually is not put on the wanted list. These data are obtained on the basis of work of technology on 10 mass actions since 2016.
The British Ministry of Internal Affairs reported that the system of face recognition for search of missing persons will be used not in real time to avoid wrong actions of police because of haste, and within the analysis of images from surveillance cameras.
The Minister of Internal Affairs Sajid Javid supported distribution of a sensing technology of persons by police, despite protests of the activists fighting for confidentiality.
In 2019 in Britain there will take place the judicial proceedings concerning legitimacy of use of the systems of face recognition.
FBI uses photos of drivers for creation of an unofficial system of face recognition
The second city in the USA prohibited the systems of face recognition
At the end of June, 2019 it became known of prohibition of the systems of face recognition in Somerville (the State of Massachusetts, the USA) where about 81 thousand people live. It is the second city after San Francisco where local state bodies and police will not be able to use the software for identification of the personality on the person.
Besides, in Somerville prohibited use of the data or evidence obtained using program systems of face recognition during criminal investigations or judicial proceedings. At the same time the local legislation does not limit use of face recognition by the state or federal law enforcement agencies, notes the VentureBeat edition.
Authors of the resolution blocking sensing technologies of persons call them "a functional equivalent of the requirement that each person always carried and showed the identity certificate with the photo". Also it is told about false positive coincidence at determination of persons of women, youth, immigrants or color people.
Prohibition was unanimously approved by council of Somerville into which 11 members enter. At the same time 9 of them created this bill therefore it is no wonder that it was adopted.
Most of representatives of city council consider that work of the systems of face recognition contradicts human rights and threatens with abuse of law enforcement agencies. Deputies, thus, agreed with arguments of the human rights activists who are afraid of total shadowing citizens. The authorities of Auckland and the State of Massachusetts in general also thought of similar bans.
Earlier American union of protection of civil liberties together with 70 other human rights organizations demanded from Amazon company to stop deliveries of sensing technologies of persons to the government and urged the Congress to impose the general ban on this practice.
How many suspects the system of face recognition in Moscow helped to delay
The police in Russia begin to use cameras with face recognition
On May 22, 2019 it became known of use by the Russian police officers of wearable cameras with face recognition function. Devices for the Ministry of Internal Affairs were developed by NtechLab company. Read more here.
NEC implements the system of face recognition revealing substitution of students at examinations at the universities
On May 20, 2019 the NEC company announced deployment of the system of face recognition in the Australian educational institutions NeoFace for the purpose of fraud prevention during examination. Read more here.
The police in London fine the pedestrians hiding persons from the systems of their recognition
In the middle of May, 2019 it became known that the London police fined the man for attempt to hide the person from the camera with the system of face recognition.
In December, 2018 law enforcement agencies in the British capital began to test the new software for face recognition. For this purpose on streets began to park vans with the corresponding equipment and warnings of monitoring.
One of pedestrians, passing by such car, decided to hide the person a collar of a sweater and to cover the head. Police officers noticed such behavior, delayed the man and demanded to show documents. The detainee was not a criminal, but to him wrote out a penalty in the amount of 90 pounds sterling (about 7500 rubles at the rate of for May 17, 2019) for "unseemly behavior".
Officers claim that when they stopped the man covering the face, that began to show aggression and to threaten. The detainee in a conversation with BBC reported that he pulled a sweater because on the street it was cold on the head, and police officers had no right to demand to show the face.
In the press service of police reported that police officers were instructed concerning people who cover the face or avoid cameras. To them were instructed to stop people "at discretion".
According to the The Independent newspaper, during testing of a system of face recognition in eight hours eight suspects were arrested. Only in three cases it was made on a program signal, told the edition the informed sources. Arrested were suspects of robberies and other offenses.
Before beginning testing of technology face recognition, the police stated that "any person who refuses scanning it will not be obligatory to be considered as suspicious".
The police caught the thief, using the photo of the famous actor in the system of face recognition
In the middle of May, 2019 it became known that the New York City Police Department (NYPD) used Woody Harrelson's photo in the program for face recognition to catch the criminal who resembled this famous actor. Read more here.
Prohibition of the systems of face recognition
In the middle of May, 2019 San Francisco became the first American city in which use of the systems of face recognition was forbidden. The city supervisory board voted for such initiative. Read more here.
The police began arrests on false signals of the systems of face recognition
On April 22, 2019 the 18-year-old student from New York Usman Bach filed a lawsuit against Apple for illegal arrest which was executed by police because of an error in the system of face recognition in one of shops of the American corporation. Read more here.
Panasonic mounted the cameras distinguishing persons and purchases in large supermarket chain
At the beginning of April, 2019 the large Japanese supermarket chain FamilyMart announced the beginning of use of a system of face recognition which allows to pay purchases without involvement of cashiers. Read more here.
It is possible to pay trips in the Chinese subway with the person now
In March, 2019 the Shenzhen subway started a payment system of trips by means of a sensing technology of persons. The technology earned from the Futian station, and for its implementation the pilot 5G network was started there. Read more here.
In the British prisons the system of face recognition for fight against a zapreshchenki broadcast appeared
At the beginning of March, 2019 it became known that in the British prisons Hull, Humber and Lindholm there was a system of face recognition intended for fight against transfer of the prohibited objects, such as mobile phones and drugs.
In 2018 the staff of prisons recorded more than 23,000 cases of the prohibited transfers that is 4,000 more, than in 2017. According to the intelligence data of Great Britain, some visitors deliver illicit goods in several prisons through the whole country. At the same time use of fingerprints and verification of documents without electronic message between prisons do not allow to identify such violators.
The technology of biometrics and face recognition tested in prisons uses the device IDScan and the software developed by the London company Facewatch. In prisons the technology of scanning of an iris of the eye of an eye developed by Tascent, the American company on biometric identification and security was also used.
In HMP Humber prison employees scanned persons of 770 visitors within six weeks to define who uses fraudulent identity documents or visits different prisoners at different times. Using new security systems, the staff of prison could reveal a number of suspects to whom after investigation it can be refused visit of the British prisons. Besides, collected data can serve as proofs at further investigations both in prison, and beyond its limits.
Nevertheless, similar use of a system of face recognition caused perturbation in some sectors of society. The British fighters for civil liberties of Big Brother Watch addressed to the European Court of Human Rights with the complaint to use by the government of Great Britain of technology of mass observation.
NEC started the ATMs allowing to withdraw money by means of recognition of the person
At the end of February, 2019 the NEC corporation announced start of the first-ever ATMs allowing to remove cash by means of face recognition. ATMs were created in cooperation with large Taiwan bank E.SUN Commercial Bank. Read more here.
IBM released base from 1 million photos of persons for training of the biometric systems
The company considers that the data used for the systems of face recognition are insufficiently various, and suggested to expand base of the analyzed images. All provided photos are supplied with tags with data of the craniofacial sizes, symmetry of the person, designations of age and a floor. Researchers of the company hope that these parts will help developers to reach bigger accuracy at face recognition.
Experts warn about a possibility of bias of artificial intelligence. Researches showed that the used technology is suitable for face recognition of white men, than women or people of other nationalities much better. Even own system of IBM was criticized. According to researchers of MIT, the IBM Watson system in 35% of cases could not identify black women though the share of errors at recognition of white men did not exceed 1%.
IBM took it into consideration and reported that the updated service of visual recognition Watson uses more various data sets for training. Nevertheless the share of errors at face recognition of black women is still quite big and makes 17%. The conducted researches guarded the public – the sensing technology of persons is used by law enforcement agencies, however at such level of errors there is a question of its expediency.
In the report of the Center of confidential data protection of law school of Georgetown University for 2016 it is said that the Afro-Americans become more often suspects, than white men, and errors in work of the police systems of face recognition can aggravate this situation.
Face recognition does not work in every second smartphone
At the beginning of January, 2019 the non-profit organization from Holland held testing of 110 models of smartphones and found out that the face recognition function used for blocking of devices does not work properly more than on every second device.
The research conducted by Consumentenbond and its international partners showed that for an unblocking 42 of the tested smartphones it is enough to have the photo of the owner of phone. Any photo, for example, received from social networks from surveillance cameras or any other method will approach.
Results of this research cause concern in users and security services. Use of the printed photo of the person of the owner are the first check of function of face recognition which is used by normal users and testers. But the main thing, is the first trick which malefactors for cracking of the smartphone, the face protected by identification will try to use before passing to more difficult attacks which include creation of masks or the 3D-printing heads of the owner of phone.
Any system of face recognition which does not pass "phototest" usually is considered useless. According to Consumentenbond, models Asus BlackBerry Huawei Lenovo LG Nokia Samsung, Sony also Xiaomi did not pass similar tests. In a case with Sony the test all models failed. Six more models - Honor and six models of LG - passed test only in the "strict" mode. Though based on this test users can conclude that you should not include face recognition, 68 devices, including the flagman Apple iPhone XR models and XS, sustained this simple attack, as well as many other high-performance models on Android from Samsung, Huawei, OnePlus and Honor.
The complete list of models which passed the phototest can be found on the website Consumentenbond.
The most popular systems of face recognition in China
By the end of 2018 China actively uses the systems of face recognition in the most different places – shops, objects of transport infrastructure and even toilets. The authorities create a national system which will allow to identify the person for several seconds. Surveillance cameras which in the country there are about 170 million are for this purpose used, and by their 2020 there have to be 570 million So on each two Chinese will fall on one camera.
According to the Business Insider edition, thanks to the big population and weak laws in the field of personal privacy police and business companies actively create TEConts using face recognition. The largest projects are listed below.
Itself Alibaba developed own systems which will be applied in the Shanghai subway to identification of passengers using their person and a voice.
The police officers monitoring security at one of the Chinese railway stations wear special sunglasses with face recognition function. The device is capable to identify the person for 100 milliseconds and more than once helped law enforcement agencies with capture of criminals.
In the Chinese Shenzhen for the first time in the world the camera of fixing of violations by pedestrians started working. She is set on one of intense transitions of the city and monitors the people who are stealing a march on an inhibit signal of the traffic light. For determination of the identity of the violator the camera uses a sensing technology of persons.
At entrance examinations in colleges through the whole country face recognition and fingerprints is used to guarantee that examined are these students.
After a number of kidnappings some kindergartens open doors only for those people whose persons are registered in a system. In one of kindergartens mounted more than 200 cameras for security.
Even in some toilets installed automatic machines with face recognition. The device issues 60 cm of toilet paper to one person not a thicket than time nine minutes.
At KFC restaurants it is possible to pay the order by means of the scanner of the person. The registered users making purchases using Alipay can use a system. To pay food the client needs to look at the three-dimensional webcam which will make the picture and will scan his face. Information tied to an account in Alipay will allow to verify the user and to make purchase.
Clients can also use face recognition for payment of purchases in the shops managed by one of the largest Chinese online retailers JD.com.
Alibaba has shops with cashless payment of Hema in which users scan the person and enter the phone number for carrying out payments via the Alipay system.
The Alibaba company together with the producer of the hotel information systems Shiji installed the system of face recognition for registration in 50 hotels. The Chinese tourists resorting to services of the Fliggy online travel agency (belonging to Alibaba) can reserve at first in it hotel, and then, using "mask" of the person quickly to become populated in hotel and to issue a deposit.
China Merchants Bank installed in China ATMs with face recognition. They use specially developed algorithm which compares the person of the person who came for money with base of questionnaires of clients of bank.
Alibaba deployed the vending machine for car sale using technology face recognition.
The Taikang insurance company checks the identity of clients on their person, using the software created by Baidu search engine.
Beijing decided to fight against illegal lease of state housing using the smart locks distinguishing owners on the person
At the end of December, 2018 it became known that in the state housing of Beijing the accelerated rates "smart" locks with a sensing technology of persons are implemented. With their help local authorities strengthen measures against an illegal repeating an examination in lease of the state housing provided to needy families on preferential quotations.
It is supposed that by the end of June, 2019 locks with the built-in system of scanning of persons will be used in all programs of providing preferential state housing in Beijing with participation of 120 thousand tenants, The South China Morning Post with reference to the Beijing edition The Beijing News reports.
Comparing information obtained when scanning persons of visitors with images from the saved database, a system will recognize owners and does not open a door for strangers, the director of the information center at the Beijing state housing center Shan Zhenyu told in Beijing News interview.
Besides, a system can be used for supervision for lonely elderly people. If the aged person during the certain time frame does not leave and does not come into the house, managing on the real estate the notification on need to come with check will be sent.
In such large megalopolises as Beijing, lease of housing very expensive. On average the rental apartment in the capital of Celestial Empire costs approximately 5 thousand yuans a month (about $730) whereas the rent for state housing can make less than 2 thousand yuans a month ($290).
The authorities of Beijing hope that the smart locks recognizing owners by the person will increase security, will prevent illegal transfer to sublease and guarantee that only really needing people use a privilege.
As of the end of 2018 smart locks with face recognition are involved in 47 programs of providing preferential state housing in Beijing. With their help about 100 thousand the scan images of persons of tenants and members of their families are received.
The Chinese Airbnb puts in houses "smart" locks with face recognition
At the end of 2018 Xiaozhu announced the beginning of implementation of "smart" locks with face recognition in houses and apartments which are suggested to rent using service. The new technology, the identity of tenants allowing to check, became a part of great efforts on improvement of protection and security in the fast-growing market of short-term lease of housing. Read more here.
Failure in London. The system of face recognition in the subway recognizes nobody
At the end of December, 2018 it became clear that the system of face recognition developed in the London subway recognizes nobody. The London police officers are criticized for use of unmarked vans for check of disputable and inexact technologies of automatic face recognition at Christmas buyers. Read more here.
Toilets with face recognition in China reduce consumption of toilet paper
At the end of 2018 it became known of the number of public toilets growing in China with the system of face recognition which allows to save toilet paper.
In December such toilet earned in Baotu Spring Park in the city of Jinan (Province of Shandong) located in 400 km to the south from Beijing. In this bathroom there is an automatic machine issuing toilet paper after scanning face. For one approach the device issues about 70 cm of paper, and for receiving an additional portion sanitary gigiyenicheskgo the same person needs to wait for a product 9 minutes and again to bring the head to the camera for identification.
It is reported that thanks to a biometric system consumption of toilet paper was reduced from 10 to 4 rolls a day. Use of a toilet and paper is free.
Theft of toilet paper and also hand soap and paper towels from public bathrooms is very widespread in China. For this reason in some toilets refused issue to visitors of toilet paper at all, and people should bring it with themselves.
Most of Chinese who tested the doser of toilet paper, say that they are not disturbed by need of scanning face. The main shortcoming, according to them, is that sometimes the issued paper appears insufficiently and therefore it is necessary to wait until the terminal is able to issue additional volume.
Equipment installation, limiting issue of toilet paper, became a part of the large-scale program which received the name "Toilet Revolution" which is implemented in China since 2015. It was started against the background of extremely deplorable status of toilets in the country and discontent of many tourists from their visit. With respect thereto the Chinese toilets had extremely bad reputation.
In 2019 in the People's Republic of China it is going to construct or repair about 21 thousand toilets on destinations. In the 2018th about 24 thousand bathrooms were open and upgraded.
For a smartphone unblocking hackers and police print the owner's head on the 3D-printer
By the end of the year 2018 many smartphones are equipped with function of an unblocking by means of recognition of the person. In December the Forbes magazine decided to check whether this technology is safe if to use 3D-printing model of the head. It turned out that hackers or police can quite use 3D-printers for gaining access to mobile devices.
The journalist of Forbes Thomas Brewster (Thomas Brewster) printed on the 3D-printer three-dimensional model of the head in Backface company in the British Birmingham for £300 (25 thousand rubles). For its production pictures from 50 cameras were used. After editing images the three-dimensional model full-scale which reminds the real person turned out.
The test was carried out on five different smartphones - iPhone X LG G7 ThinQ Samsung Galaxy S9 Samsung Galaxy Note 8 and OnePlus 6. After setup of function of an unblocking Brewster could unblock all four Android-smartphone using a 3D model of the person. He reported that all phones on Android which participated in an experiment were unblocked with "different degree of ease".
iPhone X became the only phone which did not manage to be unblocked. When passed it the presentation of Apple claimed that it worked with Hollywood studios and tested Face ID technology with 3D-printing masks and other objects to find out whether it is possible to forge them.
Face ID uses an array of sensors for user identification while Android phones is relied, as a rule, on a selfie camera which is less exact.
Sensing technologies of persons were widely adopted in the USA and Great Britain by 2018. It sparked sharp criticism from fighters for the civil rights which claim that it is necessary to enter more stringent laws for protection of the rights to personal privacy. For example, in the USA biometric data, including fingerprints and model of the person, are not protected by the fifth amendment. It means that though the police cannot force the person to enter the password, it can compulsorily apply a finger to the scanner or bring phone to the person and also create 3D - the copy of the head to unblock the gadget.
At 14 American airports the system of face recognition earned
On August 20, 2018 at 14 American airports the system of face recognition earned. The Service of customs and border control of the USA told about its efficiency ( U.S. Customs and Border Patrol, CBP).
As reported on the website of the department, on August 22 the 26-year-old passenger who arrived in the Washington airport of Dulles from São Paulo (Brazil) showed the passport of the citizen of France on control point. However a biometric system revealed that the man's face does not match the photo in the document.
When arrived in the USA sent for additional examination, he "was obviously nervous" and as it became clear, not for nothing. Found the identity certificate addressed to the citizen of the republic of Congo which actually the detainee was in his shoe. Now for attempt to drive in the USA under false documents it is threatened by imprisonment.
CBP hope that the biometric systems will allow not only to reveal on violators, but also to accelerate the procedure of passing of passport control of the passengers arriving from abroad. At the same time by August, 2018 it is allowed to citizens of the USA to refuse scanning face.
It is supposed that in the future the system of face recognition will replace boarding passes and also verifications of the passport at all stages before boarding. However the Union of protection of civil liberties ( American Civil Liberties Union, ACLU) warns that despite all promises the government can use technology for secret observation of people even if they are not suspected of anything.
The systems of face recognition of police of Britain were useless
In May, 2018 it became known of big problems in the systems of face recognition which are used by the British police officers. A large number of claims can be as a result submitted — this question became "priority" for Management of the commissioner according to (Information Commissioner's Office), cites a BBC words of the representative of the regulator Elizabeth Denham.
The British human rights organization Big Brother Watch published the results of a research which showed the "stunning" number of innocent people of whom the sensing technology of persons made potential criminals.
So, from May, 2017 to March of the 2018th a system issued for police of the Southern Wales 2685 coincidence of people to the database of suspects, however 2451 from them were false.
The London law enforcement agencies applied technology of identification of persons on a carnival Notting Hill in 2017. Indications of a system were wrong in 98% of cases when the signal that the suspect from the police database is allegedly noticed worked. The solution is arranged so that at identification of the possible violator of the law on the person on duty panel the signal comes to the nearest police station.
The police began to blame the cameras issuing the low-quality picture and the fact that a system was used for the first time, but also in the subsequent 15 actions (football matches, festivals, parades) during which involved technology, the result did not improve. Only on three a system was never mistaken.
The police also told that in nine months of system operation of face recognition she truly noted more than 2 thousand people that led to 450 arrests. At the same time nobody was included in the conclusion mistakenly. This results from the fact that in addition to work of algorithms people who check operations are involved in work and make the final decisions.
Scientists invented a new method of deception of the systems of face recognition
Every day of a system of face recognition become more difficult and are even more often used in universal life, for example, last year the company Apple let out smartphone iPhone X, equipped biometric with a system Face ID. However the similar systems it is possible to deceive, in particular, using infrared light emitting diodes. Infrared beams are not seen a simple eye, however the majority of cameras can catch infrared signals.
The Chinese researchers created The baseball cap equipped with miniature infrared light emitting diodes which are placed in such a way that the infrared beams falling on the person of the owner of a headdress help not only to hide his personality but also "issue itself for other person for passing of the authentication based on recognition of the person". This task more difficult also requires use of deep neural network for recognition of the static image of the person and the correct projection of infrared beams on the person of the impostor.
For verification of the theory researchers used photos of four casual people, they managed to deceive the systems of face recognition in 70% of cases on condition of existence of small external similarity between the victim and the impostor.
"On the basis of our finds and the attacks, we can draw a conclusion that it is difficult to call the sensing technologies of persons existing today safe and reliable in aspect of critical scenarios, such as authentication and observation", - researchers concluded. They also added that infrared light emitting diodes can be hidden not only in baseball caps, but also in umbrellas, hair or wigs.
The Russian twins demand with Apple 20 million that iPhone X does not see between them a difference
Twin brothers from Vladimir — 26-year-old Alexander and Ilya Tunchiki — sent Apple a claim to the Russian office of the company because system of face recognition Face ID on them smartphones iPhone X equally identifies both young people, thereby, according to them, breaking personal data protection.
The offended users require from the company to improve technology and also to compensate moral damage in the amount of 20 million rubles, TASS the lawyer Roman Ardykutsa representing the interests of brothers reported in January, 2018.
"Twins purchased … iPhone X to use function of an unblocking of the screen by means of persons. To their disappointment, each device recognizes both brothers what they were not warned upon purchase about, this information is absent in the instruction. For this reason applicants ask the company to finish technology" — he explained.
Face recognition in retail
In November, 2017 the CNBC TV channel released the plot telling about implementation of the systems of face recognition in shops. Retailers use such technologies for collection of data on clients and selection of offers on the basis of the relevant data.
In retail face recognition is applied generally to motivate buyers. For example, if the person is recognized on an input in shop and see its history of purchases, then the staff of shop knows better what to offer it. So, if he bought the TV in shop of electronics, the employee recognizes him, will address by name and will suggest to purchase the new panel.
According to the Hong Kong IT company Jardine One Solution (JOS), many retail networks apply possibilities of face recognition to collect data on visitors of the shops.
| ||One of things in which large traditional retailers begin to be engaged now is tracking of the one who comes into their shop and as it behaves there — the managing director of JOS Mark Lunt reported.|| |
JOS helps the retail companies with face recognition of clients for the purpose of drawing up a profile of buyers and tracking of their actions with outlet. It is about such data as the number of visitors, their age, sex, ethnic origin. Such data help shops to know better about a flow of clients and to select the personalized offers for them, noted Lant.
For example, using data analysis, arriving from the systems of face recognition, it is possible to select the music playing in a trading floor.
JOS speak, all data retrieveds of clients are anonymous, however the question of confidentiality remains relevant. Technologies do not interfere with implementation of such systems, but there are concerns connected with personal data and culture, Mark Lant recognizes.
He added that retailers spend huge amounts of money for data loss prevention and data protection. The scandal connected with data theft of millions of clients of Uber shows that the companies cannot feel safe, and users should show care, disclosing personal information, the managing director of JOS considers.
The founder and the CEO of HeadCount company (offers shops of service in monitoring and improvement of attendance) Mark Ryski says that biometric data, including those which generate the systems of face recognition belong to the category delicate and have high potential — especially for the purpose of security and improvement of customer service quality.
According to the senior vice president for maintenance strategy of clients of InMoment company Brennan Wilkie, use of the equipment for face recognition in trade premises really has a high potential. For example, such devices are capable to compare a look of the client in shop with data on him, its loyalty to a brand and other purchases. To mitigate a problem of confidentiality of users, shops need to show to clients what advantages they get as it was with cash desks of self-service or with bank cards with chips in due time, he is sure.
Earlier the Chief information officer of Dixy Vladimir Muravyov in an interview of TAdviser told about projects on targeted advertizing in shops of one of retail chain stores Victoria using a sensing technology of persons. In more detail about it here.
Authorization in iPhone X on the person was cracked with a mask for $150. Video
The Vietnamese firm on cyber security of Bkav showed a method of a bypass of a system of authorization on the person of Apple Face ID, implemented in new mobile devices of the company — iPhone X. For this purpose they needed to make on the 3D-printer a mask which only parts reminds the person of the owner of the device.
You watch Apple Face ID in more detail
How to bypass the scanner of the person on Samsung Galaxy Note 8
The web designer Mel Takhon posted on the twitter video how it is easy to bypass the scanner of the person on Galaxy Note 8. Takhon keeps two Note 8 opposite to each other on one of which — his photo in the experiment, and on another — the included system of scanning face.
You watch Samsung Galaxy Note 8 in more detail
Deception of biometric protection of Samsung Galaxy S8
The system of biometric identification in the latest Samsung Galaxy S8 smartphones it is possible to deceive using the photo, participants of the German hacker community Chaos Computer Club (CCC) found out. They told opening about the on the community site on May 22, 2017.
You watch Samsung Galaxy S8 in more detail
Allied Market Research
Allied Market Research foretells growth of the market of the systems of face recognition to 9.6 billion dollars by 2022 at average growth rate of 21.3% a year. The market leader, according to forecasts, will become the USA. 3D - technologies will occupy the most part of the market in comparison with 2D, and the market of software will grow for 23.9% annually till 2022. The market will increase including due to distribution of applications in the field of face recognition.
Researchers learned to deceive the system of face recognition
Face recognition is based on neural networks which study at a set of examples. Show to the computer thousands and millions of persons, and it begins "understand" where a nose, eyebrows and so on. After that, if to show a certain face, then the computer reads out patterns known to him; analyzes what the request is similar to; also issues the answer.
Scientists from Carnegie's University — Mellon learned to deceive the system of recognition. For this purpose they colored with points in such a way that the computer began to be mistaken and took the person for another. For testing they used already existing neural networks which persons are able to distinguish very well — sometimes better.
Researchers managed to issue the white man for Milla Jovovich almost in 90 percent of cases. The computer in so many percent of cases took the woman of Asian appearance in special glasses for the man from the Middle East.
Besides, they tried the method on the commercial Face program ++ which is used in Alibaba for payment authorization. In this case they did not put the person wearing spectacles in front of the camera, and at first took his picture wearing spectacles and then loaded it into the program. As a result they managed to issue one person for another in 100 percent of cases.
Public organizations of the USA against face recognition
The coalition from 52 public and human rights organizations of the USA sent to the Ministry of Justice the letter with a request to investigate overutilization of sensing technologies of persons in work of law enforcement bodies. Also the coalition is disturbed by the unequal accuracy of machine face recognition of different race which can become a basis for manifestation of racism from staff of bodies.
Especially the local police, police of states and FBI abuses these technologies, the letter says. The coalition asks the Ministry of Justice to be engaged first of all in check of those police departments which already are under examination in connection with the prejudiced attitude towards citizens with not white skin color.
Results of a research of the Center of privacy and technologies of School of the right of the university of Georgetown formed the basis for a request. The research showed that persons of a half of adult population of the USA under different circumstances were scanned by government identification software.
Researchers note that in the USA there are no serious rules regulating use of this software today. According to Alvaro Bedoyi (Alvaro Bedoya), the director of the Center and the coauthor of a research, having photographed on driver's licenses, the person already gets to base of persons of police or FBI. It is especially essential taking into account that face recognition is inaccurate, and in this case can do harm to innocent citizens.
Examples of projects in HSBC, MasterCard and Facebook
The HSBC bank brings together portrait gallery of the clients. The financial conglomerate passes to the new system of identification — a selfie. The photo will replace all different ways of determination of the personality as, for example, fingerprints, recognition of a voice and introduction of the PIN code.
The service will be available to corporate clients of HSBC. Through banking mobile application they will be able to open accounts on one click of a selfie. The bank confirms the client's personality using the program of face recognition. The photo is compared to the pictures which are earlier loaded into a system, for example, from the passport or driver's licenses. It is supposed that new service will save from need to remember digital codes and will reduce identification time.
MasterCard announced in the spring of 2016 on the international congress of mobile Mobile World Congress technologies in Barcelona that it will soon permit to use self as an alternative for passwords at online payments. Service will be available next summer in the USA, Canada and some countries of Europe, such as Italy, France, the Netherlands, Great Britain and Spain.
To use this option, users will need to download a special application on the computer, the tablet or the smartphone. Then to look in the camera or to use the device scanner for fingerprint recognition (if it is available on the device). However (at least, at the moment), users still will need to provide data of the bank card in addition. Only if additional identification will be required, users will be able to use the above described option.
Thanks to such new approach, MasterCard is going to protect users from counterfeit online transactions which are performed using stolen passwords of users and also to provide to users more convenient system of authorization. The company reported that 92% of people who tested this new system preferred it to traditional passwords.
Some experts doubt data protection from that cyber-criminals could not receive easily fingerprints of the user or the photo of his face if transaction is performed at unsafe use of public network Wi-fi.
Experts in cyber security claim that a system should include several security levels for prevention of potential theft of photos of the person of users. Online payments represent an attractive target for cyber-criminals.
At the end of 2015 the group of experts from the Technical University of Berlin showed a possibility of extraction of the PIN code of any smartphone when using a selfa of the user. For this purpose they read out this code which was displayed in the opinion of the user when it entered it on the OPPO N1 phone. It is rather simple to hacker to intercept control over the front camera of the smartphone for accomplishment of this elementary attack. The cyber-criminal could intercept control of the user's device, make his self and after that execute online payments using the typed password which the hacker saw in the opinion of the victim?
MasterCard insists that its mechanisms of security will be able to detect similar behavior. For example, users will need to blink for the application to show a "live" image of the person, but not his photo or previously shot video. A system compares the image of the person of the user, converting it into the code and transferring him under the safe protocol via the Internet in MasterCard. The company promises that this information will be safely stored on its servers, at the same time the company will not be able to reconstruct the user's person.
In the summer of 2016 it became known that researchers bypassed the system of biometric authentication, using a photo from Facebook. The attack became possible thanks to the potential vulnerabilities inherent in social resources.
Team of researchers from North Carolina State University showed a method of a bypass of the security systems constructed on a sensing technology of persons by means of available photos of users of social networks. As it is explained in the report of specialists, the attack became possible thanks to the potential vulnerabilities inherent in social resources.
"It is not surprising that the personal photos posted in social networks can pose a confidentiality threat. Most large social networks recommend to users to set settings of confidentiality at the publication of a photo on the website, however many of these pictures are often available to general public or can be browsed only by friends. Besides, users cannot independently control availability of the photos placed by other subscribers", - scientists note.
Within the experiment researchers selected photos of 20 volunteers (users of Facebook, Google+, LinkedIn and other social resources). Then they used these pictures for creation of three-dimensional models of persons, "recovered" them using a number of animation effects, imposed texture of skin on model and modified a look (if necessary). Researchers tested the turned-out models on five security systems, four of them managed to be deceived in 55-85% of cases.
According to the report of Technavo company (winter of 216) of one of the key trends having a positive impact on the market of technologies of biometric identification for the person (facial recognition) implementation of the multimodal biometric systems in such sectors as health care, the banking, financial sector, the sector of securities and insurance, the sector of transportations, the motor transport and also in a public sector is.
The multimodal biometric systems constructed on a combination of several biometric technologies, such as fingerprint recognition, features, a voice, etc. differ in high efficiency of detection of unauthorized access to devices of bank self-service, to databases of a health care system, mobile devices and also a large number of on-line and offline applications.
Due to the growing need for increased security for Europe steady growth of use of the systems of biometric identification on the person is expected. As of 2015, despite the fact that Europe is the second largest participant of the world market of technologies of biometric identification for the person other technologies, such as fingerprint recognition, the drawing of veins on a hand and an eye iris of the eye, are widespread more widely. Implementation of the facial recognition systems was performed by lower rates that was connected with crisis in the eurozone. But analysts expect that within the next four years cumulative rates of annual growth of this market will exceed 21%.
Producers invest considerable means in scientific research and development of systems of biometric identification on the person. It is expected that it considerably will accelerate development of such systems due to identification of qualitative parameters of the person, including scars, lengths of a nose or a look and which can be used for determination of age or a sex of the person.
Technologies of biometric identification for the person can be used in the field of retail for identification of clients and tracking of their purchases, purchasing habits, age, sex, crime and credit story. It is expected that the data received using such systems will be used by retailers in the marketing purposes and to do to clients special offers on the basis of information on their previous purchases.
At the final stage the Internet buyer will need to make purchases the photo using the smartphone. MasterCard believes that it is much easier, than to remember passwords.
According to CNN Money, using the new innovation tool the payment system is going to reduce fraud level. "I think, it will be pleasant to new generation which lives pictures of a selfie. They for certain will pick up this technology", – Ajay Bhalla, the director of MasterCard of the innovative solutions in the field of security reported.
MasterCard uses a security technology of online payments of SecureCode which assumes introduction of the password for payment confirmation on the Internet. According to the company, this technology was used in 3 billion transactions for last year, it prevents cases use by swindlers of the card on the Internet. However passwords are forgotten, can steal or intercept them. For this reason many financial companies began to implement biometric technologies for convenience of users and increased security.
At first the project will cover 500 users, and in case of successful testing – will be started for public use.
2014: Creators of the Israeli startup of IsItYou are going to use front cameras of smartphones as a method of confirmation of the identity of people at commission of bank transactions. According to data of the edition of the Air Force, Israelis are sure that they in future selfie will be able to replace passwords, fingerprints and other forms of identification of the personality. IsItYou implemented a new sensing technology of persons, the having fine precision and protection against fraud.
The founder of the project Benjamin Levy told that he thanks to the high level of security of IsItYou will be able to distinguish 99999 from 100 thousand cases of deception. Levy tried to convince banks about need of implementation of its system next year. It will be used for carrying out financial transactions.
Google already uses function of recognition of the person in Android. Thus it is possible to unblock the device running this mobile OS. Nevertheless, developers repeatedly claimed that recognition of the person is insufficiently protected in comparison with classical methods. With respect thereto experts began to doubt in Binyamin Levi's statements.
Marios Savvedes from Carnegie Mellon University researches function of recognition of the person. He considers that IsItYou which is independently tested on security cannot be reliable.
The world expert in the field of biometrics doctor Massimo Tistarelli holds the same opinion. He told that in Europe the full-scale scientific Tabula Rasa project which main goal - development of protection against fraud for biometric methods of identification is carried out. According to him, before entry into the market it is necessary to conduct a number of the independent researches confirming efficiency of a product.
- Directory of solutions and cybersecurity projects - Biometric identification
- Biometric market today
- Biometric identification (market of Russia)
- National Biometric Platform (NBP)
- Biometric identification (world market)
- Technologies of biometric identification
- Myths of biometrics
- Short history of biometrics
- Biometrics and border control
- Biometric systems of accounting of working time
- Video surveillance systems
- Video analytics
- Authentication systems
- Cybersecurity - Fraud detection system (fraud)
- ↑ of the Sensing technology of persons or a face control rationally
- ↑ Chinese murder suspect ‘caught by AI software that spotted dead person’s face’
- ↑ Facial recognition to be used by UK police to find missing people
- ↑ Boston suburb bans facial recognition software
- ↑ Moment man is fined of £90 for hiding face from police facial recognition cameras
- ↑ Prison Facial Recognition Technology Could Deter Smuggling
- ↑ IBM hopes 1 million faces will help fight bias in facial recognition
- ↑ Facial recognition does not work as intended on 42 of 110 tested smartphones
- ↑ Gezichtsherkenning op smartphone niet altijd veilig
- ↑ 22 eerie photos show how China uses facial recognition to track its citizens as they travel, shop — and even use toilet paper
- ↑ Beijing turns to facial recognition to combat public housing abuses
- ↑ China’s public toilets have facial recognition now … thanks to Xi Jinping
- ↑ We Broke Into A Bunch Of Android Phones With A 3D-Printed Head
- ↑ Facial Recognition Technology Catches Imposter at Airport, Officials Say
- ↑ Face recognition police tools 'staggeringly inaccurate'
- ↑ Scientists invented a new method of deception of the systems of face recognition
- ↑ Invisible Mask:
- ↑ to Practical Attacks on Face Recognition with Infrared
- ↑ the Russian twins demand with Apple 20 million that iPhone X does not see between them a difference
- ↑ Facial recognition is tracking customers as they shop in stores, tech company says
- ↑ Authorization in iPhone X on the person cracked with a mask for $150. Video
- ↑ Accessorize to a Crime: Real and Stealthy Attacks on State-of-the-Art Face Recognition
- ↑ than the person Kak to descend for Milla Jovovich in the opinion of the computer Researchers learned to deceive the system of face recognition
- ↑ the System of face recognition of FBI accused of racism
- ↑ of HSBC will distinguish clients on a selfie
- ↑ of MasterCard suggests to use a selfie instead of passwords
- ↑ the Israeli developers offer biometric identification by means of a selfie