The Moscow government has announced another tender - worth 2 billion rubles. - to supply high-performance equipment for video analysis and storage of 1.5 petabytes of information from the capital’s cameras. Today, 175 thousand cameras are operating in the capital, in 2020 they will be 25 thousand more.
It is assumed that the “smart” technique will not only help to catch criminals, prevent offenses, but also contribute to the replenishment of the city’s treasury with fines. At the same time, various monitoring and analysis systems will be integrated into a single network, which will be replenished with new elements. For these purposes, new technologies will continue to be introduced in transport, stadiums, and in crowded places. However, while experts doubt that such a system will work efficiently: there are no similar successful projects anywhere in the world.
In November last year, NefAZ (a subsidiary of KamAZ) and LiAZ agreed to equip 490 buses purchased by Mosgortrans with Internet of things systems, including monitoring passenger traffic. The city has bought "Stuffed" buses for 7 billion rubles. All buses will be transferred to the city until September 30, 2020.
The MaximaTelecom company participates in the project, which built Wi-Fi networks in the Moscow and St. Petersburg metro, and together with partners organized a unified Wi-Fi network in the capital for passengers of land transport and the MT_FREE metro. Another project participant is Itelma, a manufacturer of electric transport equipment. According to Kartoteka.ru, NPP Itelma LLC is owned by Ilya Zak (45%), Alexander Postnikov (38.7%), Leonid Zakulin (15.8%) and Yuri Filin (0.5%). An expert estimate of the revenue of suppliers of IoT solutions for Mosgortrans buses in this project will amount to 2.5 billion rubles.
In 2020, sensors also equip 1.5 thousand metropolitan buses of private carriers. While we are talking only about buses, in the future it is possible that solutions will be distributed to other modes of transport.
From January 1 to December 26, 2019, in the capital's public transport (metro, buses, trams, trolleybuses), controllers identified 578 thousand fare dodgers and fined them a thousand rubles each. It is assumed that the new equipment will recount the birds with one stone, and fare dodgers will be able to bring the capital treasury much more than 600 million rubles received by the controllers.
Door open sensors will provide information on the number of incoming and outgoing passengers, and video cameras will “check” travel documents and identify fare dodgers.
Clever technology is already responsible for calculating the passenger flow in transport, but it is often mistaken, because it is still poorly analyzing information. For example, CCTV systems with face recognition available on the market do not take into account people who simply let passengers out. For some solutions, the error can reach up to 100%. Svyatoslav Irishin, leading manager of the IT Systems for Public Transport area of EuroMobile, believes that “malfunctions often occur at rush hour, when people gather near the sensors. Another way to count the “hares” is the smart step: the pad on the floor of the vehicle, which responds to pressure from the passenger’s leg. But she is not faultless. "
The expert believes that here devices with the analysis of video information can be useful: “Devices with the analysis of video“ lead ”each individual person. If the passenger let out other people and walked away not far (about a meter), then the sensor does not consider him a "hare." However, counting accuracy above 95% is guaranteed by units of manufacturers of video devices. Most decisions have an average error of 20–30%. ”
Through the eyes of the cops
If the large-scale introduction of systems with in-depth analysis in transport is still a matter of the future, the metro and significant urban facilities can already report on the results. In the subway, smart cameras are looking not only for stowaways, but also for other offenders. Devices identify faces and verify them with wanted bases. Upon coincidence, police officers immediately receive a notification.
It is difficult to describe the operation of the system in detail, since neither DIT, nor the Ministry of Internal Affairs, nor Deptrans chose to answer such questions. You can only rely on the stories of the offenders themselves, who walked past cameras with face recognition.
The system was described by Mikhail Agel, an activist of the Other Russia organization. Agel and his friends went down the escalator at the Sportivnaya metro station. A policeman came up to him, asked for documents, then pointed to the display of a smartphone or similar device. The open application displayed a photograph of a person taken by a camera with face recognition and a brief reference: name, passport data, for what actions are wanted. Photos were taken by a camera with face recognition at the time of the descent to the Sportivnaya, and the system immediately recognized the criminal in it. However, Axel’s questionnaire did not include the case number, investigator, or other important information. It turned out that the National Bolsheviks passed according to the guidelines as a “football hooligan”, but instead of the “administrator” he was given the status of a felon. A few minutes later he was released.
read all text
According to the data of the Main Directorate of the Ministry of Internal Affairs, 100 people were detained in the metro using VisionLabs solutions (included in the ecosystem of Sberbank; conducts research work in the field of computer vision, data analysis and robotics) in 2018, of which 62 were wanted. VisionLabs did not provide relevant information, citing the fact that they act as an executor in the project.
According to “Ko” in the Main Directorate of the Ministry of Internal Affairs in Moscow, with the help of 1 thousand cameras with face recognition technology at the entrances of residential buildings within the law enforcement segment of the APK “Safe City”, 205 offenders were identified for 11 months of 2019, in 2018 there were fewer offenders - 119 people.
The department did not talk about how the operational-search actions occur after the detection of offenders at the entrances of the houses.
“Until September 1, in full, this system (face recognition) will be introduced in the subway. This means that wanted persons will be identified in the subway in a fraction of a second, ”said Moscow’s mayor Sergei Sobyanin at an expanded meeting of the board of the Main Directorate of the Ministry of Internal Affairs of Russia in Moscow at the end of January 2020.
Stadiums won't hold bullies
Face recognition technologies are also used to ensure safety in stadiums. The Moscow Department of Information Technology previously reported 152 wanted individuals who were detained with cameras at public events from 2018 to June 2019.
For the 2018 FIFA World Cup, turnstiles, ticket readers and video identification systems were installed at Russian stadiums. The latter integrated with the databases of the Russian Football Premier League. As a result, the entrance to the stadiums was closed to citizens blacklisted by clubs and the Ministry of Internal Affairs (by the way, the list itself is entered upon entry into force of a court order).
“The analytical modules of video surveillance will identify the attacker. The alarm is then sent to the operator or security service for further action. In integrated systems, cameras transmit “commands” directly to the checkpoint or launch other protection devices (for example, smoke alarm systems). In this way, an attacker’s hit on an object is minimized, ”explained Anton Golubev, director of project management at Hikvision Russia & CIS.
Similar algorithms are also used in retail chains, airports, banks and other facilities. Face recognition modules simultaneously identify up to 100 faces in a frame. “The database is not limited, and information is easily added from other sources,” said Rustam Salimzibarov, director of development at Macroscop.
However, it is too early to talk about the large-scale introduction of technology. This system was not involved during the Moscow protests, said Mikhail Klimarev, executive director of the Internet Protection Society. “Judging by the fact that everyone was caught in a row, no“ face recognition ”was used. This technology is not regulated by Russian legislation, there are no laws and regulations, ”said the expert.
Big problems of the "big brother"
Now 175 thousand surveillance cameras are operating in Moscow: in courtyards, in entrances, in parks, and buildings of public institutions. Of these, 3 thousand recognize faces.
All information flows into the Unified Data Storage and Processing Center (ECHR). In courts, records are used as evidence. The plans of the Ministry of Internal Affairs and city authorities include increasing the number of cameras to 200 thousand and equipping them with face recognition function.
Experts say that the introduction of the new system is already accompanied by problems. Thus, federal legislation does not directly regulate facial recognition systems.
“Processing biometrics without the written consent of citizens violates the law on personal data and the right to privacy (Articles 23 and 24 of the Constitution of the Russian Federation), and also contradicts the principle of the presumption of innocence,” said Anton Ershov, chairman of the Russian Pirate Party.
Mikhail Klimarev, executive director of the Internet Protection Society, agrees with him: “Power departments, as long as there is a regulatory gap, can do anything. It seems to me that a problem has ripened in society on the application of this technology and an open discussion is needed.
The problem with data leakage remains unresolved. Access to cameras is sold on forums, and people working with the ECCH are doing this, according to the MBH Media. You can connect to the live broadcast of a specific camera, see the archive for five days. Full access to the ECSD, according to journalists, costs from 30 thousand rubles.
“Cameras with face recognition help to learn more about the behavior of a particular person and his environment, control movement around the city. The data will be “useful” to attackers for blackmail or robbery, ”said Mikhail Klimarev.
Another problem is the fragmentation of databases on offenders. It is unlikely that the city’s cameras will recognize a visiting criminal. “The Safe City system in Perm does not work with the Safe City system in Yekaterinburg. For the system to work effectively, data exchange between state and municipal departments, state and commercial structures and between competitors is necessary, ”said Rustam Salimzibarov.
Not surprisingly, the number of opponents of face recognition technologies and biometric “anarchists” is growing. They are in the social network "VKontakte". The largest community, “For the abolition of the collection of biometrics, DNA, chipization, 5G”, has 13.8 thousand people.
The scientific community does not trust face recognition. Some decisions define African Americans and Asians 10–100 times less accurately than Caucasians, researchers at the US National Institute of Standards and Technology (NIST) found.
For a combination of reasons, the European Union plans to suspend face recognition projects in public places for up to five years. Officials discuss the terms of the ban and develop rules that would protect citizens' personal data from leaks.
In Russia, representatives of the expert community give only recommendations on “face conservation”. “The easiest is to partially hide the face, for example with a scarf. A more elegant way is to apply a special make-up of lines and patterns on the face, this is misleading to most face recognition algorithms. Glasses with infrared LEDs that illuminate the camera will also come in handy, ”explained Anton Ershov. These methods still work. But the market has already announced the release of a solution for recognizing people by silhouette.
China did not digest the information
The introduction of a face recognition system for 200 thousand cameras is a technically difficult procedure, explained Mikhail Klimarev. Each camera for acceptable facial recognition accuracy should have a resolution of at least 720 pixels, a video recording camera generates a minimum of 16 gigabytes per day. 200 thousand cameras per day will record about 3 petabytes of information, the expert noted. According to him, processing such a volume of data will require appropriate hardware and software. “In addition, face recognition is an extremely costly procedure in terms of computation,” he explained.
According to Mikhail Klimarev, total face recognition in such volumes is technically difficult to implement. Even China has failed in this task of national importance. “A huge amount of money is allocated from the PRC budget for hardware and software. But the desire of the authorities to organize total surveillance in some autonomous regions was not crowned with success, ”the expert noted.