Step 1
Rules on the frame
The head of security marks the fence line, the loading dock zone and the checkpoints on the cameras and sets the time: “Perimeter — around the clock”, “Dock — from 22:00 to 6:00”.
Video analytics for site security
Security cameras record around the clock, but one duty officer watches them. NeuroVision video analytics spots intrusion over the perimeter, an abandoned object, a crowd and a car at the gate, and immediately sends the event with a frame. The system only suggests a match with a list — the decision is made by the guard.
Security video analytics
Your cameras and video server The decision is the guard’s A cloud in Russia or your own perimeter
Archive retention period — clause 30 of the anti-terrorism protection requirements for places of mass gathering (Russian Government Resolution No. 272 of 25.03.2015)
Video analytics for security: the officer sees an event, not forty screens
A site has dozens of cameras and one or two people at the post. Attention to the screen drops after half an hour, and the recording is opened only after something has happened. Video analytics for site security watches all the cameras at once and shows the officer only what needs a response.
The cameras are already there — as required
Many sites must run video surveillance and keep an archive. The analytics works on top of those cameras.
- 01
An alarm with a picture
A sensor says “triggered”; a camera with analytics shows what exactly: a person climbed over the fence near warehouse 4 at 02:17. The officer knows at once who to send and where.
- 02
An archive you can search
Instead of watching a night at 16× speed — a list of events: every crossing of the line, every car at the gate, every abandoned object. Reviewing an incident takes minutes.
- 24 hours no fatigue at the monitors
- 0 cameras to replace to get started
- 1 month free pilot
What video analytics notices on a guarded site
Site security video analytics works with rules you draw on the frame: a zone, a line, a time and who to notify. Five of the six events do not require identifying a person.
-
Intrusion over the perimeter
A person crossed the fence line, climbed over the fence or entered a closed zone after closing. Security receives an event with a frame, not a sensor signal without a picture.
-
Abandoned object
A bag, box or suitcase stays in a zone without an owner longer than the set time — at an entrance, in a lobby, on a platform. The officer sees the frame and the moment the object was left.
-
Crowding
More people have gathered in a zone than you set: at a shopping centre entrance, in an evacuation corridor, at the turnstiles. The event arrives before a queue turns into a crush.
-
Loitering at a door and night movement
A person stands at the door of a server room or a cash desk longer than usual, someone walks around the warehouse after closing. The rule runs on a schedule — during the day it does not interfere with work.
-
Vehicle plates at the entrance
A camera at the checkpoint reads the car’s plate and checks it against a list: your own vehicles, contractors, today’s guests. The entry log keeps itself; an unknown car is an event for security.
-
Match against a list of faces
A face in the frame resembles a person from your list — the guard gets the frame and the photo side by side and decides whether it is the same person. More — in the block below.
How the officer handles an event: from the alarm to the log entry
Example — the Talnikovy logistics centre: 64 cameras, a perimeter with two checkpoints, a night security shift of three people.
Step 2
An event at the post
At night a person crosses the fence line near warehouse 4. The officer’s screen shows an event: the frame, the “Perimeter-11” camera, the time 02:17 and a short video clip.
Step 3
Response and check
The officer sends a patrol and looks at the neighbouring cameras. If it is a night-shift employee who took a shortcut, the event is marked “false” with a comment.
Step 4
Log and review
In the morning the head of security sees the shift log: events, responses, marks. Repeated false alarms are a reason to adjust the zone or the schedule.
Warehouse, shopping centre, place of mass gathering: security scenarios
-
Warehouses, industrial sites, logistics
Perimeter and fence, loading docks at night, vehicle plates at the checkpoint, movement on the site after closing.
-
Shopping and business centres
Abandoned objects in lobbies, crowding at entrances and escalators, service areas and doors after closing, parking and entry.
-
Places of mass gathering and city facilities
Railway stations, parks, squares, sports and concert venues. Crowding, abandoned objects, events in zones; city video analytics works on top of the cameras the site already runs as required.
-
Parking lots and restricted areas
Entry by a list of plates, a car staying longer than allowed, a person in a restricted zone at night.
The set of events depends on the site. For industrial and warehouse infrastructure the perimeter matters, for retail and city sites — people and objects.
Hard hats, vests and danger zones in manufacturing — on the page video analytics for manufacturing. Staff access by face — Face-recognition access control.
Anti-terrorism protection and cameras: what Government resolutions require
The anti-terrorism protection requirements are set by the Government — separately for each type of site. Below is a summary of two resolutions that mention video surveillance directly, and of the rules on faces in the frame. It is a guide as of October 2026, not a legal opinion: the requirements for your site are in its security passport.
- The requirements for the anti-terrorism protection of sites and the forms of security passports are approved by the Government (clause 4, part 2, Art. 5 of the Federal Law “On Countering Terrorism”)
- Different sites have different resolutions: places of mass gathering, retail sites, hotels, sports and educational facilities and others. Which requirements apply to you is shown by the categorisation act and the security passport
- Three categories by the number of people who may be at the place at the same time: over 1,000, from 200 to 1,000 and from 50 to 200 people (clause 10 of the requirements, Government Resolution No. 272 of 25.03.2015)
- Whatever the category, the place is equipped with video surveillance, alert and evacuation control, and lighting systems (clause 23)
- Video surveillance is continuous, across the whole territory; the archive is kept for 30 days; the system interacts with the regional video surveillance subsystems of the “Safe City” (clause 30)
- A retail site, whatever its category, is equipped with video surveillance, alert and evacuation control, and lighting systems (clause 30 of the requirements, Government Resolution No. 1273 of 19.10.2017)
- Video surveillance is continuous, across the site’s territory, with archiving and data retention (clause 31)
- The requirements speak of video surveillance and an archive; analytics adds real-time events and archive search to the same cameras
- Events and clips go to your video surveillance system or via API; its link to the “Safe City” is set up by you and your integrator
- Biometric personal data is information from which an operator establishes identity (part 1, Art. 11 of 152‑FZ); it is processed with written consent, except in cases named directly in the law, including legislation on countering terrorism and on transport security (part 2, Art. 11)
- Organisations may not collect and store biometrics in their own systems for identification, except collection for the UBS (clause 1, part 1, Art. 15 of 572‑FZ); automatic identification goes through the UBS (clause 2, part 1, Art. 15)
- 572‑FZ does not apply to a check with an authorised official taking part (clause 2, part 2, Art. 1) — that is why a list match is confirmed by the guard
- The retention period for the archive and the entry log, who has access and where the cameras are placed are set in a local act; for places of mass gathering — at least 30 days (clause 30 of Resolution No. 272)
- The archive and events are stored in the cloud on servers in Russia or on your own servers with no internet access; access is by role
- Resolution No. 272 — as amended on 24.10.2023; Resolution No. 1273 — as amended on 04.03.2026; 152‑FZ — version of 26.07.2026; 572‑FZ — version of 28.12.2024. What the UBS is — a blog article.
A list match: a hint for the guard, the decision is theirs
The face is the only thing in security video analytics that counts as biometrics. So the system does not decide on its own here: it shows the guard a candidate, and a person compares and decides.
-
You compile the list
The list holds only people you have a legal basis for: consent or a case named directly in the law. Who adds people, who sees the list and how long it is kept, you decide as the data operator.
Basis:
part 1–2, Art. 11 of 152‑FZ
-
A candidate, not a verdict
When a face looks similar, the guard gets the camera frame and the photo from the list side by side on the screen. The system does not open a door, does not call a patrol and does not write “identified”.
Decision:
the guard on duty
-
The guard confirms or rejects
The guard compares the frame and the photo themselves, goes up and checks documents if needed. Their decision and the time are recorded in the log.
Basis:
clause 2, part 2, Art. 1 of 572‑FZ
-
Automatic entry — through access control
If you need the turnstile to open by itself, without a guard, that is a different scenario — turnstiles with face recognition with a Liveness check.
Learn more:
Face access control
Connecting to video surveillance, the guard post and access control
- Your cameras and archive
- Cloud in Russia or perimeter
- 1-month pilot
- Events via API
Terms
Pilot — 1 month free
Cloud in Russia or your own server without internet
Your cameras and archive
How much video analytics for site security costs
The price depends on the number of cameras, the set of events and where processing runs. Start with a free pilot on the perimeter or in a lobby.
-
Several cameras and an archive of past nights, marking of lines and zones, review of events with your security team.
- Calculate the cloud
Cloud
on request
Processing on servers in Russia, events on the officer’s screen, a log, API.
- Calculate installation
In your perimeter
on request
Installed next to the video server, with no internet access, a licence by the number of cameras.
Cameras, the video server, barriers and installation are not included in the price. Video analytics is not in the price matrix — we calculate the cost for your site.
Site security glossary: perimeter, abandoned object, category and security passport
Terms from the protection requirements, the contract with a security company and a conversation with an integrator.
- Perimeter
- The boundary of the guarded territory: a fence, a barrier, a line on the camera frame.
- perimeter
- Line crossing
- An event when a person or a vehicle crosses a line set on the frame.
- line crossing
- Abandoned object
- An object that stays in a zone without an owner longer than the set time.
- abandoned object
- Crowding
- The number of people in a zone above the threshold set for it.
- crowd detection
- Plate recognition
- Reading a vehicle’s plate in the frame and checking it against a list.
- ANPR / LPR
- Place of mass gathering
- A territory or public place where, under certain conditions, more than 50 people may be at the same time (clause 6, Art. 3 of 35‑FZ); the requirements for its protection are in Government Resolution No. 272.
- public gathering place
- Site category
- The protection level that a commission assigns to a site during a survey; the mandatory measures depend on it.
- site category
- Security passport
- A site’s document with its category and anti-terrorism protection measures.
- security passport
- City video analytics
- Analytics on the cameras of streets, squares, transport and places of mass gathering.
- smart city video analytics
- List match
- A frame that resembles a photo from your list; the decision whether it is the same person is made by the guard.
- watchlist match
- On-premise
- Installing video analytics on the customer’s servers, with no video leaving the perimeter.
- on-premise
Licences and certificates
The project «Development of a multimodal system using artificial intelligence "Neuromed" for managing the diagnostic processes of a clinic on the basis of the "Neurovision" computer vision platform» is supported by the Innovation Promotion Foundation within the federal project «Domestic Solutions»
The Neurovision computer vision platform is registered with Rospatent (certificate No. 2023661508) and included in the Russian Software Registry (entry No. 19972 of 13.11.2023)
The right to use the Platform is granted under a licence agreement concluded by Neurovision with its customers
We will set up video analytics for your guarded site
Tell us what the site is, how many cameras there are and which video surveillance system, which events matter more — perimeter, objects, people or entry, whether you need a server with no internet — and we will offer a pilot on your recordings.
Can't put your request into words? Book a call and our team will contact you at a convenient time
Book a call
The company is included in the register of IT companies and the register of personal data operators of the Russian Ministry of Digital Development.
Address: Moscow, Presnenskaya Embankment 12 — Moscow City business centre, Federation Tower. Nearest metro stations: Delovoy Tsentr and Mezhdunarodnaya.
Read also
Materials on face recognition, algorithm accuracy and protection against deepfakes — in our blog.
542 29
KYC, AML and Digital Identification
How Video Streams Are Spoofed in KYC: The Architecture of Attacks via Emulators, Virtual Cameras, and Video Stream Injection
187 18
KYC, AML and Digital Identification
Unified Biometric System (EBS): What It Is, How to Enroll Your Biometrics, and Where It’s Used
404 16
KYC, AML and Digital Identification
How to ensure the protection of personal and biometric data in KYC processes
174 23
KYC, AML and Digital Identification
Face Recognition System: How It Works, Its Accuracy, and Where It’s Used
Questions about video analytics for security
A neural network that analyses video from security cameras and marks events itself: intrusion over the perimeter, an abandoned object, crowding, a car at the gate. The officer receives an event with a frame instead of watching every screen in turn.
No. It notices the event, and security responds: sends a patrol, checks, records the result. One officer with analytics can follow many more cameras.
You set the zones — a lobby, an entrance, a platform — and the time after which an object without an owner counts as abandoned. The system shows the frame and the moment the object appeared, and the officer decides what to do.
Yes. A camera at the entrance reads the plate and checks it against a list of your vehicles, contractors and guests. The entry log keeps itself; its retention period is set by your local act.
Only as a hint against your list: the guard sees the frame and the photo from the list side by side and decides whether it is the same person. The system cannot and will not automatically recognise everyone; automatic entry by face is an access control scenario.
Yes. Under Government Resolution No. 272 such places are equipped with continuous video surveillance with a 30-day archive. The analytics works on top of those cameras: crowding, abandoned objects, events in zones — and search of the same archive.
Usually not. Installed IP cameras are suitable; for plates you need a camera that sees the plate close up, for the perimeter — a camera with a view of the fence line. We will choose during the pilot.
The requirements for places of mass gathering provide for the site’s video surveillance system to interact with the regional “Safe City” subsystems. NeuroVision passes events to your video surveillance system or via API; the connection to the city segment is set up by you and your integrator.
In the cloud on servers in Russia or on your own servers with no internet access — for sites where video must not leave the perimeter. Access to events and the archive is separated by role.
The first month of the pilot is free. After that we calculate by the number of cameras and events, separately for the cloud and for installation in your perimeter.
