The Moscow video surveillance system, which the authorities use to search for criminals and oppositionists, uses four facial recognition algorithms at once. The BBC Russian Service writes about this.
According to the BBC, in mid-summer, the Electronic Moscow company, subordinate to the Moscow City Hall, published four contracts for services for «data indexing» in the «urban segment of the unified video analytics system» for a total amount of about 800 million rubles — 200 million for each. Contracts were concluded with a single supplier, without competition and publication of any documentation.
Thanks to the published data, journalists found out exactly what technologies the Moscow authorities use. Three of them turned out to be Russian — NtechLab (in which Rostec owns a stake), Tevian FaceSDK and VisionLabs Luna Platform. Another one is Kipod, a product of the Russian-Belarusian company Synesis. It was this algorithm that helped the Belarusian security forces find opponents of the Lukashenka regime during the suppression of the 2020 protests. One of the founders of Synesis, Alexander Shatrov, was on the US and EU sanctions list. «alt=»1» />«Impossible to hide.» How an activist of the «Other Russia» was detained in the metro on a signal from a video camera that identified him by the orientation of the Center «E»
According to the ByPOL, with the help of cameras connected to Kipod, the Belarusian security forces were able to track down and detain opposition blogger Nikolai Dedok.
On its website, Synesis confirms that Kipod is being used for surveillance at Moscow metro stations. In addition to searching by photo, according to the developers, she can find a person by signs — you need to specify his gender, race, approximate age, whether he wears glasses, a beard or mustache.
The VisionLabs Luna Platform algorithm is used by several dozen Russian banks. For example, Alfa-Bank, thanks to her, checks a potential client in the “black lists” of credit bureaus even before he asks for anything and submits his documents.
Luna, like Kipod, according to the developers, determines the age, gender, race and even emotions of a person. This is the official partner of the Moscow City Hall in the Face Pay project. As part of it, metro passengers can upload their selfies to a special mobile application, link a bank card to their profile, and pass through the turnstiles with one glance at the camera.
Tetian, which developed the FaceSDK algorithm, was founded by people from the faculty's multimedia laboratory Computational Mathematics and Cybernetics, Moscow State University. This system can also detect a person's emotions.
As the BBC was told in the Department of Information Technology of Moscow, all four algorithms are used by the authorities at the same time.
“Video streams from cameras are simultaneously processed by all four algorithms. Then, according to a special formula, the result of the work of all algorithms in the aggregate is calculated, ”the Moscow Department of Information Technology reported.
Also, according to the BBC, the algorithms used in Moscow compete with each other on accuracy and speed.
The Moscow mayor's office has almost 190 different information systems, with the help of which it collects and analyzes information about citizens, houses, cars, stalls and playgrounds.
For several years now, work has been underway on a single system that will be able to extract data from all resources and platforms of the Moscow government. The authorities also tried to create the IBD-F 2.0 system, which would accumulate data from various sources, at the federal level. It was supposed to replace the outdated «IBD-F», which unites the police bases of all regions of Russia.
«IBD-F 2.0» was planned to be connected to the databases of other government departments — up to information about travel by public transport and biometric data. The state contract, according to the BBC, was concluded in 2020, but was terminated due to «non-fulfillment of conditions.»
According to journalists, the Ministry of Internal Affairs has other systems. For example, the Moscow police use the «Accounting and Barrier System — M», which allows you to conduct an end-to-end search through arrays of other bases and form checklists of citizens of operational interest. And the city video surveillance system connected to the facial recognition system helps to find them.
With the help of it, the police on the Russian Flag Day, August 22, detained participants in past anti-war actions. Footage from the city's video surveillance system is also in the FSB video about Natalya Vovk, the alleged killer of Alexander Dugin's daughter Darya. On the footage of the security forces, Vovk enters the entrance of a Moscow house, the system marks her as “alien”. This means that she was not previously seen among the inhabitants of this house.