AI Security Systems Auckland | Facial Recognition & ANPR NZ

Q: What is AI surveillance and how is it different from standard CCTV?

Standard CCTV records video and relies on a person to review it. AI surveillance analyses the video stream continuously, detecting defined events such as a person crossing a line or a face matching a watch list, and generates real-time alerts. You find out about incidents as they happen rather than after the fact.

Q: What is ANPR and how does it work?

ANPR uses a dedicated camera with an AI engine to read vehicle number plates, log them with timestamps, and compare against white lists or black lists. Matched plates trigger alerts or automated outputs such as opening a boom gate.

Q: Is AI surveillance legal in New Zealand?

Yes, subject to the Privacy Act 2020. Facial recognition involves biometric information treated as sensitive personal data, requiring documented purpose, access controls, and retention limits. Comsys provides a written commissioning pack for every facial recognition install.

Q: What is the difference between AI analytics and traditional motion detection?

Traditional motion detection triggers on any pixel change, including passing headlights, shadows, or foliage. AI analytics distinguishes between object classes and event types before generating an alert, resulting in far fewer false alarms and more meaningful notifications.

What AI surveillance actually does differently

Traditional CCTV records and waits. AI surveillance analyses in real time, filtering the noise and surfacing the events that actually matter. Instead of reviewing 16 hours of footage to find one incident, you get a timestamped alert the moment it happens. Instead of generic motion detection that fires on every passing car, you get a targeted notification when a person crosses a defined line at 2am on a Tuesday.

Every Comsys AI security install begins with a free site assessment. We identify what you need the system to detect, which cameras require AI analytics capability, and how alerts should route — to your phone, to a monitoring station, or to a security desk. The result is a system that works as a genuine operational tool, not a recording archive you only access after something has already gone wrong.

Facial recognition — identify persons of interest, create watch lists, and receive alerts on detection

ANPR (Automatic Number Plate Recognition) — capture and log every vehicle plate at gates, driveways, and entry lanes

AI behavioural analytics — detect loitering, running, fighting, and other behavioural anomalies

Crowd monitoring and density alerts — for retail, events, and public-access buildings

Line crossing and intrusion detection — virtual tripwires and perimeter zones with real-time alerts

Object removal and left-object detection — alert when items are removed from or left in defined areas

Request an AI security quote

ANPR and facial recognition in detail

Automatic Number Plate Recognition (ANPR) is the most commonly requested AI feature for Auckland commercial and industrial sites. A dedicated ANPR camera is positioned at a gate, driveway entry, or car park lane. The on-camera or NVR AI reads each plate as vehicles pass, logs the plate and timestamp, and compares it against a white list (authorised vehicles) or black list (flagged plates). Matched plates trigger an instant alert or can be used to automate gate or boom-gate opening for authorised vehicles.

Facial recognition operates similarly: enrolled faces are compared against the video stream from selected cameras in real time. When a face matches a watch list entry, an alert is generated with a cropped image and timestamp. This is used in retail for known shoplifters, in corporate settings for access audit, and in high-security environments for identity verification at entry points. Privacy Act 2020 obligations apply, and Comsys includes a written data-handling record in every facial recognition commissioning pack.

ANPR accuracy — Hikvision and Dahua ANPR cameras achieve 95%+ plate reads in well-lit conditions; we assess lighting during the site visit

Watch list management — add, remove, and update plates or faces from the NVR or cloud management portal

Multi-lane ANPR — separate dedicated cameras per entry lane for high-traffic sites

Integration with access control — ANPR can trigger relay outputs to open boom gates or intercom panels automatically

Privacy Act compliance documentation — included in every facial recognition installation

Book a site assessment

Smart perimeter protection and behavioural analytics

Smart perimeter protection uses AI to distinguish a person or vehicle crossing a boundary from a branch moving in the wind, a cat walking along a fence, or a headlight sweeping across the frame. The practical result is far fewer false alarms and far more reliable alerts when something genuinely warrants attention. Hikvision AcuSense and Dahua WizMind are the platforms we use most; both run the AI processing on the camera or NVR directly, so no cloud subscription is required for analytics to function.

Behavioural analytics extends this further. Line crossing detection creates a virtual tripwire — anyone crossing from outside to inside (or in either direction) after hours triggers an immediate alert. Intrusion detection creates a zone where any person present triggers an alert after a configurable dwell time. Object removal detection flags when a fixed object (a piece of equipment, a vehicle, a display item) leaves its defined position. Running detection can trigger a crowd-control response in a retail environment. These are not future capabilities — they are standard features on current Hikvision and Dahua IP cameras and NVRs that Comsys installs and configures today.

Line crossing detection — virtual tripwires with directional control and time-based arming

Intrusion zone detection — configurable dwell time before alert to avoid false positives

Object removal and left-object detection — for retail floor, reception, and equipment areas

Loitering and running detection — for car parks, retail environments, and school grounds

No cloud subscription required — analytics run on-camera or on-NVR

Discuss your site requirements

Frequently Asked Questions

AI security systems — common questions

What is AI surveillance and how is it different from standard CCTV?

Standard CCTV records video and relies on you (or a monitoring operator) to watch it. AI surveillance analyses the video stream continuously and automatically, detecting defined events — a person crossing a line, a face matching a watch list, a plate matching a database — and generating an alert in real time. The practical difference is that you find out about incidents as they happen rather than after the fact, and the system ignores the vast majority of routine activity that standard motion detection would flag as an alarm.

How accurate is facial recognition for security purposes?

Modern facial recognition on current Hikvision and Dahua hardware achieves recognition accuracy above 95% in well-lit conditions with frontal face presentation. Accuracy drops in poor lighting, with heavily obscured faces (hats, masks), and at very long distances. We assess camera placement and lighting during the site visit and will advise honestly where facial recognition is and isn’t going to give reliable results for your specific layout.

What is ANPR and how does it work?

ANPR (Automatic Number Plate Recognition) uses a dedicated camera with an embedded or NVR-based AI engine to read vehicle number plates as vehicles pass the camera field of view. The system captures the plate characters, logs them with a timestamp, and compares them against configured white lists (authorised) or black lists (flagged). Matched plates trigger alerts or automated outputs — such as opening a boom gate for authorised vehicles. The camera must be positioned correctly for the approach angle, lighting, and vehicle speed; Comsys designs the camera position during the site visit.

Is AI surveillance legal in New Zealand?

Yes, subject to the Privacy Act 2020. CCTV surveillance (including AI analytics) on private property is legal when its purpose is proportionate, when signage notifies people of recording in public-facing areas, and when footage is retained for a defensible period and not disclosed inappropriately. Facial recognition specifically involves biometric information, which the Privacy Act treats as sensitive personal information requiring documented purpose, access controls, and retention limits. Comsys provides a written commissioning pack for every facial recognition install that documents the data-handling arrangement to support Privacy Act compliance.

What is the difference between AI analytics and traditional motion detection?

Traditional motion detection triggers on any pixel change in a defined area — passing headlights, moving shadows, insects flying past the lens, or wind-blown foliage all trigger alerts. AI analytics distinguishes between object classes (person, vehicle, animal) and event types (crossing a line, loitering, running) before generating an alert. In practice this means a well-configured AI system generates a fraction of the false alerts a traditional motion-detection system produces, making alerts meaningful rather than something you learn to ignore.

AI security systems for Auckland