FAMICA Vision · AI CCTV
Fewer cameras. Fewer alerts. Better evidence.
AI CCTV for malls and commercial buildings — placed by a coverage audit, filtered at the edge, and explained by one platform your guards and your property team both use.
The operating reality
Cameras see everything. Teams cannot.
An ageing analog estate
Low resolution, coax of unknown condition, DVRs reviewed only after the event. Every added camera is another screen nobody watches.
Blind spots where it matters
Car parks, loading docks, back-of-house corridors, rooftop plant — the highest liability and the fewest eyes.
Alarm fatigue
Motion alerts fire on weather, animals, shadows and signage. Within weeks the control room learns to ignore them.
Evidence without context
The footage exists. Finding it, linking it to the guard's response and proving the timeline takes a day.
Phase 0 · Coverage audit
The honest way to end up with fewer cameras.
AI does not reduce camera count. Resolution, field of view and placement do. Before any hardware is quoted we walk the site and produce a zone-by-zone plan: what each analog camera covers today, what a correctly placed AI camera would cover, and what the estate looks like after.
Plans are illustrative. The real numbers come from the site walk, camera by camera — that is the point of the audit, and why it comes before any quotation.
What you get
A zone map, a before/after camera schedule, cabling and network findings, priority geo-fences, and a phased plan that reuses what is still sound.
Phased, not rip-and-replace
Where coax is in good condition, Ethernet-over-coax lets an AI IP camera use the existing cable. Racks, power and recording positions stay where they make sense.
Acceptance criteria up front
Alerts per day, response time, cameras per zone, false-alert rate. Agreed before the pilot, measured during it, reported at the end.
At the edge
Detect people and vehicles — not weather.
The camera does the first job on its own chip: a person from a swaying tree, a vehicle from a shadow, a loiterer from someone walking through. Only credible events leave the lens.
| Rule | Where it earns its keep |
|---|---|
| Geo-fence & intrusion | Loading bays after hours, rooftop and M&E rooms, back-of-house stairs |
| Line crossing | Service corridors, fire exits used as shortcuts, tenant boundaries |
| Loitering & dwell | ATM lobbies, car-park stairwells, unattended entrances |
| People & vehicle counting | Entrances, ramps, dock turnaround |
| Low-light performance | Basements and perimeters where analog shows grey noise |
- Specified by risk, not by brand. From the established AI-camera families — Dahua WizSense, Hikvision AcuSense, Uniview LightHunter — on coverage, environment and lifecycle cost.
- Supplied and installed by a certified partner, warranted by the manufacturer. Detection analytics are the camera makers' own.
- Cameras are never the lock-in. The platform above the recorder is where the value lives, and it works across makers.
- No facial recognition by default. See privacy below.
Above the recorder
What FAMICA adds to a camera.
An AI camera makes an alert. FAMICA makes it an incident with an owner, a response and a record — and turns the same footage into numbers the property team can use.
Event hub & escalation
Credible events become incidents routed to the right guard's phone — verify, dispatch or dismiss in one tap. Nothing fires twice; nothing is lost.
Incident timeline & audit
Clip, zone, who was notified, who responded, how long it took — a defensible record for insurers, tenants and management.
Ask the footage
"Every loading-dock vehicle after 10 pm last week." Natural-language search across events, and an incident report drafted for sign-off.
Footfall & dwell
Corridor and entrance heatmaps by hour and day, from cameras you already needed for security.
Car park & dock
Occupancy by level, entry and exit counts, dock turnaround time — operations metrics from the perimeter cameras.
Sensors on the same timeline
A door contact, a leak probe and a camera event in one view. One building, one picture, one team.
Architecture
From detection to coordinated action.
- AI cameras
Edge detection on the camera. Person, vehicle, zone, line. Nuisance filtered before it leaves the lens.
- NVR
Recording and retention on site. Evidence stays where it is captured.
- FAMICA Core
Rules, incidents, escalation, search, analytics, audit — and the API into patrol, incident and CAFM systems.
- Guards · control room · property team
Mobile for the guard, dashboard for the control room, reports for management.
Open architecture: existing recorders and patrol systems are integrated, not discarded. Phased modernisation, zone by zone.
Privacy by design
People and vehicles — not faces.
A mall processes personal data the moment a camera is switched on. FAMICA Vision is built so that a compliance review is short.
- Detection, not identification. Rules act on "a person entered a zone", never on who. No facial recognition unless a client specifically requires it and completes its own impact assessment.
- Retention you set. Footage stays on the site recorder under your policy. FAMICA stores events, clips tied to incidents, and aggregated counts.
- Role-based access and a full audit log. Who viewed what, when and why — exportable for a PDPA request or an internal review.
- Analytics are aggregate. Heatmaps and counts never carry an individual's track outside the incident it belongs to.
- Signage and policy support. We supply the notice wording and the data-flow description your DPO will ask for.
The pilot
From risk map to operating rhythm.
- 0
Coverage audit
Site walk, zone map, before/after camera schedule, cabling findings.
- 1
Fit
Cameras and recorder to the plan and the budget; the partner installs.
- 2
Calibrate
Two to four weeks of rules and thresholds tuned on real traffic.
- 3
Enable
Guards on mobile, control room on dashboard, escalation paths tested live.
- 4
Report
Results against the acceptance criteria, and the plan for the next buildings.
A pilot should end with a number — alerts per day, minutes to respond, cameras per zone — not with an opinion.