Home Use Cases City Control Centres
USE CASES · MUNICIPAL CONTROL CENTRES

City Control Centres

“An incident just came in. How many hours does finding that person take you today?”

Operators scrub through days of footage by eye. And after a 4K upgrade, that footage is sometimes already gone. UniverseAI solves both without replacing a single camera — we keep the footage, and we make the footage searchable with a few conditions.

Send Your Site Conditions Request a Live Demo

What actually happens in the control room

We will not open with market size or growth rates. Here are the three scenes your team lives through every time.

SCENE 1 / REWINDING
Days of footage, reviewed by eye
An incident comes in. You have a description of the person, but no way to search by it — so an operator watches days of recordings from the beginning.
Cause: the footage is stored, but it is not organised into anything you can search.
SCENE 2 / FOOTAGE ALREADY GONE
The footage for that date no longer exists
Months after the 4K upgrade project closes, an investigation request arrives — and the footage for that date is already gone. It is not an equipment failure; it becomes an accountability problem for your team.
Cause: retention days = storage capacity ÷ daily video volume. Raising image quality multiplies the denominator, and unless storage grows with it, retention shortens by the same factor.
SCENE 3 / AFTER YOU FIND THEM
Route and companions are still manual work
You have found the person. But where they passed through and who they were with still means opening camera after camera by hand.
Cause: a search returns one scene — route and companions are a separate job.
These three scenes look like separate problems, but they sit on one line — footage must survive for an index to accumulate, the index must accumulate for search to work, and search must work before route and companions are possible. If the first link breaks, none of the rest holds.

AI BOX + VCA — retention and search on one line

Neither your cameras nor your existing VMS get replaced. We insert one stage in between, and layer search on top.

A configuration diagram in which the existing cameras and the existing storage and control systems stay as they are, and only two stages — AI BOX and VCA — are newly inserted Camera → storage → control room. We insert just two stages into that flow. Grey = exactly what you use today / blue and navy = the new stages Unchanged New stage Unchanged New stage Unchanged Existing cameras Ordinary CCTV No replacement AI BOX On-site compression + face index One box = 4 channels Storage Existing VMS No replacement VCA Condition-based search Companion network Movement trail Control operator Reviews candidates Makes the final call Retention → index → search → trail. Each stage stands only if the one before it does.

Only two stages are new. The cameras in front and the storage and control systems in the middle stay exactly as they are.

Component What it does Why it is needed
AI BOX Next to the camera, it reduces video volume by roughly 85% (based on field measurement) and builds a face index while compressing. One box covers four channels. You keep your retention policy even after moving to 4K. Search only exists if the footage survives.
VCA Combines face, body, clothing colour and carried-item conditions to narrow candidates, then shows companion relationships and movement routes. Turns days of rewinding into a handful of candidates. It layers onto your existing cameras as they are.
Used together Retention, indexing, search and route connect into a single line. It removes the state where retention succeeds but usage fails.
“Retention is about keeping the footage. Indexing is about making that footage actually usable.”
4K at 16Mbps, as is
7.5 days

On the same storage capacity

Codec improvement only (approx. −50%)
15 days

Still short of the common 30-day policy

With AI BOX (approx. −85%)
50 days

Clears the 30-day line with margin

Thirty days is not a legally fixed number — it is the common operating policy of public agencies. Send us your site conditions and we will calculate against them.

Adoption takes four steps

From confirming site conditions to procurement — here is exactly what is exchanged at each step.

1
Confirm site conditions
Channel count, storage capacity, retention policy in days, and image-quality baseline including any 4K upgrade plan. With those four we calculate how many days your current setup actually reaches.
2
Live demo
We run condition-combination search, companion mapping and movement routes live. For compression, the original and the compressed stream are shown side by side for you to judge yourself.
3
Pilot verification
Verification under the same conditions as the real site. Scope and duration are agreed case by case.
4
Procurement
The sales partner is the procuring party; UniverseAI supplies the product and the technical basis — measured data, integration specifications and retention-day calculations.
About the procurement track
Public procurement includes a track where passing a designation programme allows the contract to proceed by direct negotiation. Because such programmes and their applicability change often, we do not assert specifics on this page — the procedure is guided by the sales partner and the responsible department. Budget is normally allocated under a control-centre upgrade programme, not as consumable or component unit pricing.

What we base these claims on

Only what has been measured and what has been deployed — followed by the limits, stated just as plainly.

99.97%
Face recognition rate under KISA certification testing
Under 1 sec
1:N matching response
400M+
Processing records
1,000 channels
Live government city-scale deployment
Four certifications

KISA, iBeta, GS Grade 1 and ISO. If you need them in a specific procurement document format, we prepare them accordingly.

Origin and deployment

A Korea-developed, non-Chinese vendor. Deployable both on-premises and in the cloud, including environments where no data may leave the premises.

Stated up front
  • We narrow the candidates; your officer makes the final determination. This is not a feature that automatically identifies a person from a single photograph.
  • Control-centre CCTV is distant, angled downward, and mixes night-time and backlit conditions. Figures measured under optimal conditions do not reproduce as-is in the field.
  • It works with most ordinary cameras, but accuracy drops on cameras with very low image quality or poor angles. We confirm the minimum image-quality baseline (standard 2MP-class CCTV) with you in advance.
  • Compressed quality is not “lossless” — it is a level at which degradation is hard to perceive, and you compare it against the original side by side in the demo.
  • Real-time alerting (event detection) is outside the scope of this configuration. Tell us the situation you need covered and we will review the requirements and respond.

Frequently asked questions

The nine questions officers actually ask first.

Do we have to replace all our cameras?
No. We use the ordinary cameras already installed, integrating via ONVIF and RTSP. That said, cameras with very low image quality or poor angles can reduce accuracy, so we confirm the minimum image-quality baseline in advance. Send us your current camera list and we will review it and respond.
What happens to the VMS we already use?
It is not replaced — we layer on top of it. Tell us the name and version of your VMS and we will confirm the integration method and respond.
If we upload one photo, does it automatically identify the person?
It is more accurate to see it as narrowing candidates. We reduce days of rewinding to a handful of candidates, and your officer makes the final determination.
If the angle does not capture the face well, is the person unfindable?
Even without a usable face, you can search on conditions such as body, clothing colour and carried items. In the field this combination is used more often. You select and combine conditions, and we run it live in the demo.
Can we still meet a 30-day retention policy after moving to 4K?
Give us four things — channel count, storage capacity, retention policy in days, and image-quality baseline — and we will calculate it. For reference, at 4K 16Mbps a codec improvement alone reaches only about 15 days, which falls short, while on-site compression raises it to around 50 days.
Does video data leave our premises? Our policy does not allow cloud.
Both on-premises and cloud deployments are possible, and in an on-premises configuration the data stays within your environment. Once the deployment form is fixed, we document the exact data flow for you.
How are personal data and audit requirements handled?
We provide access-rights management and audit logging, and operating policy is configured to your institution's regulations. Tell us which regulations apply to you and we will list the required items and respond.
How does procurement proceed?
The sales partner is the procuring party and we supply the product and its technical basis. The procurement procedure and contract method are guided by the partner and the responsible department.
What does adoption cost?
We respond with a quotation. Four inputs are required to prepare it: channel count, storage capacity, retention policy in days, and current image quality together with any 4K upgrade plan.

Related products

Detail pages for the two products that make up this package.

AI BOX — Edge compression device →

The retention-collapse diagram, specifications and FAQ
VCA — The video search brain →

Condition-combination search on face, body, colour and carried items, plus companions and routes

So the footage from the day of the incident is still there

Send us four things — channel count, storage capacity, retention policy in days and image-quality baseline — and we will calculate how many days your current setup reaches, then respond with a demo schedule.

Send Your Site Conditions