A face in the crowd: the FSB is watching you!

President Medvedev has made much of Russia’s need for modernisation and advanced technology. One project piloted in some Moscow metro stations involves face recognition using biometric technology. This can clearly be used as protection against terrorism, but given that the organisation which commissioned the project is the FSB, information gained could also be used for other purposes, say Andrei Soldatov and Irina Borogan

With each year that passes biometric technology, designed to recognise people by their physical and behavioural characteristics, becomes more sophisticated and less expensive, and today it is exploited with great enthusiasm by the state.

Amongst the current projects of the Commission for the Modernisation and Technological Development of the Russian Economy, created by Dmitry Medvedev in 2009 and hailed by forward minded citizens as a tool for reform, is one focussing on perfecting personal identification systems. The project is being carried out by the security services, and its aim is to create multibiometric systems for identifying individuals in real time. In practice this means technology that allows a video camera to pick out your face in a crowd, compare it with a database and determine whether you are a criminal, or even establish your identity.  

Who calls the tune? 

Strangely enough, it was the FSB, and not the MVD (Interior Ministry), that pioneered the use of sophisticated identification technology for tracking criminals many years ago.

The Modernisation Commission was set up to promote
reform and development in Russia. Nevertheless, given the
current active levels of FSB engagement in its projects,
this noble goal is probably unattainable. Photo: kremlin.ru

At the Commission for Modernisation, responsibility for the development of biometric systems is in the hands of Working Group No 4, ‘Strategic Computer Technology and Software’, headed by Andrey Fursenko. But if you look at records of group meetings, it becomes clear that all proposals on the subject come from FSB representatives. For example, on 8 October 2009, when two projects were discussed – one the creation of an automated video system for detection and identification of targets in real time and the other concerned with voice recognition – the group was addressed by Yevgeny Maximov, deputy head of the FSB’s research establishment. Responsibility for both projects was given to the FSB and its director Alexander Bortnikov.

Initially the question was formulated thus: Russia needs its own dynamic automated human identification system using ‘the texture and three dimensional forms of the facial surface’. In practice the task was understood more widely: the FSB invited tenders for systems that would use CCTV to pick out a potential criminal by his or her walk and facial expression, and combinations of factors revealing a person’s stress levels, and even included in their specification a programme module which would be capable of distinguishing between a living face and a mask or cast.

According to Russian Federal Government Order No 13- r, dated 11 January 2011, 151 million roubles were allocated for 2011, and 157 million for 2012, for the provision of ‘video surveillance, automated video detection and identification of targets and emergency situations in real time and the creation of a real time database of identified targets’. The recipient of the funds is identified as the FSB.  

The specific customer within the security service is described as Military Unit 68240. This number masks the identity of the infamous OTU, the Technical Operations Directorate, which is responsible for wiretapping. Unit 68240’s project to develop video surveillance and identification involves seven different competitive tenders, details of which are publicly accessible thanks to legislation on government procurement.

'151 million roubles were allocated for 2011, and 157 million for 2012, for the provision of ‘video surveillance, automated video detection and identification of targets and emergency situations in real time and the creation of a real time database of identified targets’. The recipient of the funds is identified as the FSB.' 

Yevgeny Yakovlev is the technical director of Byterg, a company that in 2010 won a tender from the FSB to investigate ‘means of creating a super-high-performance reconfigured digital platform for the development of imaging and identification of targets’. I am sitting with him in a very attractive two-storey detached house on Bolshaya Pochtovaya Street. Yakovlev, a fit-looking engineer of around fifty, refuses flatly to discuss the order he has received from the FSB:

‘Yes, we’ll be completing it soon. When? Soon. No, I can’t tell you what specific targets the system is designed to identify.’

‘Could you at least say whether it’s a person, an animal, an object?’

‘I don’t know. We don’t have any contact with the customer. Our job is to comply with the technical specification: we don’t know what it will be used for.’

‘You’re not in contact or you don’t want to talk about it?’

‘Of course we’re in contact.’ 

Office of Byterg. The company is not only a leader among
the Russian producers of CCTV cameras; it is also licensed
for secret government business and is now working on the
target identification project for the FSB.

Byterg specialises in CCTV cameras. Yakovlev tells me that their factory in Aleksandrovo  turns out about 100,000 units a year. Byterg has a couple of serious Russian competitors, but for the moment most of the market is controlled by foreign manufacturers. The company was set up by some engineers in 1997, and is licensed for secret government work. It even has its own research department, which is where the FSB order is being handled.

Yakovlev only shows signs of animation when he talks about the cameras themselves. An analogue camera, he says, can only identify up to five faces in a crowd, an IP (Internet protocol) camera about 20. For identification it is also necessary to allow for the fact that the angle of divergence between the camera image and the image stored in the database must not be greater than 10-15 degrees. That is why they use security frames at meetings: as someone passes through, they cannot avoid being caught by the camera. The same tactic can be used in the Metro, and Yakovlev tells me there was a proposal for cameras to be installed beside ticket turnstiles. In fact I discovered that they have been piloting this system in the underground for some months.  

Let’s meet up at Okhotny Ryad    

In the entrance hall of Okhotny Ryad metro station ‘intelligent’ cameras have been installed; images taken by them are sent to the Metro system’s situation room, the FSB and the Interior and Emergencies Ministries. The cameras are linked to ‘Sova-Videopotok’ software, which matches biometric identification of each person caught on camera against Interior Ministry search databases.

Aleksandr Abashin, CEO of Ladakom-Service, the company that developed the system, tells me it is so advanced that a scan of 10 million images would take no more than seven seconds. This opportunity has not so far been exploited in Russia, but he intends to change all that.

Abashin is a tall, sturdily built Military Intelligence veteran. He served for most of his life as a naval officer, but he has spent the last ten years installing biometric identification systems at airports, railway stations and stadiums, and has become a real identification freak.

His office is in an old detached house on the former royal Ismailovo estate, on the only island in Moscow, reached by a bridge across the Serebryanka river. The windows look out on towers and churches surrounded by century old oaks and lime trees. In this haven of tranquillity a couple of dozen programmers spend their days perfecting the technology of identification.  

Aleksandr Kulyashov, the technical director of Ladakom-Service, with a ‘Sova-Videopotok’ mobile surveillance pack. This mobile system allows for greater flexibility and can be set up virtually anywhere in the city.

‘To put it simply’, explains Abashin, ‘the face on the photograph is measured using 30 identifiers, and the resulting mathematical matrix is very difficult to cheat. The thing about our ‘Sova’ data search system is that it was initially developed for the police, and was based, not on a reference photo, but on a composite image.’ The Interior Ministry has had a ‘Sova’ system for several years.

Abashin is very happy to tell me about the system installed at Okhotny Ryad.

‘The first thing we did was to compile online files on people being actively sought by the police or who represented some kind of operational interest, and set up a video image library for further processing. The system includes CCTV cameras with biometric identification capability installed in the northern entrance hall of the station, which send data to the 1st District police station and the Metro’s situation room.’

'A mobile system can be set up at important control points – national borders, entry points at football matches, airports, railway stations, Metro stations and so on. If we upload databases on people being actively sought by the police, we can pinpoint a person’s whereabouts at a given moment with an accuracy of up to 96%.'

Aleksandr Abashin, CEO of Ladakom-Service

Altogether the station has about 60 cameras. Abashin explains that passengers enter the frame of an Axis biometric camera as they step on to an escalator. The video footage is sent immediately to a number of authorities, including the FSB. I asked him how long this data is kept.

‘We insist that it be kept for 30 days; that is long enough for a crime investigation. The system initially sends an alarm signal to the operator’s monitor, and then the police activate their matching matrix.’

‘What percentage of people entering the station are observed by the system?’

‘Not more than 20%.’

Abashin, however, is disappointed that the Okhotny Ryad project is not being extended to other stations.

‘We are ready to continue, but there is no interest. Also, many of the metro stations in Moscow, especially in the centre, are listed buildings; we can’t do anything that would alter their appearance, so for example we can’t install cameras in the turnstiles, which would be an ideal place for them.’

The identification system does not, however, have to be fixed or installed on a permanent basis. In Abashin’s office there is a small black case containing a notebook and scanner, with a camera on a tripod next to it.

‘Look, this is a 3-in-1 unit. It’s a mobile system that can be set up at important control points – national borders, entry points at football matches, airports, railway stations, Metro stations and so on. If we upload databases on people being actively sought by the police, we can pinpoint a person’s whereabouts at a given moment with an accuracy of up to 96%. The system instantly sends an audio impulse to an operator’s workstation, a private security firm, a police unit where there is a monitor screen. 

‘How much does this unit cost and how long is its working life?’

‘A 3-in-1 system [i.e. one that can work with photo and video and also check ID documents – author] like this one costs about two million roubles. Our systems have a lifespan of more than three years; we use a reinforced Panasonic computer that can survive being dropped on the floor.’

Irina Borogan testing the surveillance system, which successfully ‘verified’ the image recorded by the camera with her passport photo. It also produced a list of the closest matches which included Irina's sister (whose photo can be seen at bottom left).

We tested the ‘Sova-Videopotok’ mobile surveillance pack on one of the authors of this article, Irina Borogan. Aleksandr Kulyashov, the technical director of Ladakom-Service, set up some lighting and led Irina towards the digital camera on the tripod. Then he scanned Irina’s passport. As a further check we also decided to scan the passport of Irina’s younger sister, since they are often mistaken for one another.

Irina’s face appeared on the monitor, in a frame and with the pupils of her eyes glowing with green light, part of the biometric matrix – the system was measuring the space between them (one of the key identifiers). After a few moments the image recorded by the camera was ‘verified’ with Irina’s passport photo.

There were, however, problems with her sister.

‘The system hasn’t verified you with your sister’s photo ’, says Kulyashov, ‘so it can tell you apart, although you are very alike. There were also photographs of men amongst those identified as resembling you most closely, but don’t worry: the system is set up at present to react to any similarity, even if it’s only a mole in the same place on your cheek. If we set up a system to control access to a building, for example, we specify a similarity coefficient of 95%, and it only takes someone to party too much at New Year and the system won’t recognise him and won’t let him through. But here it has been set up at 15% similarity.’

In the end the system did find Irina’s sister’s photo, but placed it only 5th on its similarity scale. Aleksandr explained that the rules require an operator who has received a signal to check the 10 closest matches, and after that everything depends on his decisions.

The FSB is watching you

The use of fixed biometric identification systems in Moscow’s public areas is still a rarity, but the FSB intends to change that situation in the near future. The security service has been authorised to develop an urban CCTV system incorporating a network of thousands of cameras installed in large cities on streets and squares, in airports and railway stations, and in public transport. Last year the FSB commissioned a research study on the subject, entitled ‘Defining Approaches to the Development of Regional and Urban CCTV systems’. The study was scheduled to run from November 2010 to November 2011, at a cost of 15 million roubles.

CCTV networks can already be found on the streets of
Russia as part of the 'Safe City' scheme. The current
goal of the FSB is to build up the system to enable it to
identify individuals and create a database of their images.

Video surveillance systems did not use to fall into the FSB’s remit. Since the middle of the 2000s CCTV networks have been set up in many Russian cities, as part of the ‘Safe City’ scheme. Installed by the city council, they consist of analogue cameras, which can monitor a general situation, but cannot provide high quality identification.

The FSB has set new goals: the emphasis is now on identifying individuals and building up a database of high quality video images of Russian citizens. The new CCTV systems will incorporate digital cameras with facilities for ‘intelligent video’, in other words, where the video surveillance system automatically performs an analysis of the captured video and compares it with images held in a database.

'Since the FSB is calling the tune, we can rest assured that the footage captured by the urban CCTV systems will be sent to the service’s own server and that it will have access to the systems.'

The FSB’s tactical-technical objective makes it clear that these CCTV networks in cities will be monitoring residential areas and public transport networks, as well as keeping tabs on roads and automatically recording traffic incidents. One objective is singled out for attention: ‘to monitor the situation in areas of mass public assembly, including entertainment venues, and transmission of information to the appropriate services’.

All this implies the active use of video analysis and the development of a methodology for searching for people using information from databases, in other words the tried and tested, both in Russia and the West, technology of facial recognition (as used, for example, by Facebook, to caption your party photos). The FSB even asked its designers to create a ‘mnemonic scenario description language’, which would describe the behaviour of a potential offender (also video analysis) and send a signal to the system operator.

And of course, since the FSB is calling the tune, we can rest assured that the footage captured by the urban CCTV systems will be sent to the service’s own server and that it will have access to the systems.

The project is likely to come in on schedule. According to the Russian Federation Government Analysis Centre, which monitored projects initiated by the presidential Commission for Modernisation, all FSB projects between November 2010 and April 2011 were completed successfully.

About the authors

Irina Borogan is a Russian investigative journalist who covers the operations of Russian security services. She is co-founder of the web site Agentura.Ru, which chronicles the services’ activities. Last year, Borogan and Andrei Soldatov co-authored The New Nobility: The Restoration of Russia’s Security State and the Enduring Legacy of the KGB (PublicAffairs).

 

Andrei Soldatov is a Russian security services expert, and together with Irina Borogan, co-founder of the Agentura.Ru web site. Last year, Soldatov and Borogan co-authored The New Nobility: The Restoration of Russia’s Security State and the Enduring Legacy of the KGB (PublicAffairs).