A UniFi Protect camera with license plate or face recognition runs that analysis on the camera itself and matches it against a list of plates or faces you enroll, producing a labeled event rather than a generic alert. The recognition only exists on cameras built for it — Ubiquiti's AI series and G6 series — and the processing and footage stay on your local recorder rather than a vendor's cloud. What the camera actually classifies is narrower than "it recognizes my car": it reads a plate string and compares it to a list, or it matches a face to an enrolled profile. Below is what the feature does, what it needs to work reliably, and what to ask your installer before you lean on it for daily convenience.
What does UniFi license plate and face recognition actually detect?
UniFi Protect's AI runs locally on the camera hardware itself. According to Ubiquiti, "UniFi provides advanced AI-powered detections to facilitate camera security," listing motion, person, vehicle, face recognition, and license plate recognition as the core categories, and specifying that face and license plate recognition "requires an AI or G6 series camera" (UniFi Physical Security - Ubiquiti). That's a meaningful limit: an older or lower-tier camera on an estate's network — a doorbell, a basic indoor dome, a legacy model — will not perform this recognition even if the NVR software is current. The capability lives in the camera's own silicon, not purely in the recorder.
Rather than assume every camera in a product line carries every recognition feature, ask whoever is designing your system to confirm, model by model, that the exact camera being mounted at a gate or entry position is one of the AI-series or G6-series cameras Ubiquiti specifies for recognition — not just a similar-looking camera from the same family.
How does a camera learn which plates or faces to recognize?
Ubiquiti describes the workflow as labeling: "Identify and label faces with AI series cameras to elevate your site's security," paired with "Powerful Vehicle & Audio Classification. Easily find vehicles by type and color, and filter incidents by audio events" (UniFi Physical Security - Ubiquiti). In practice this means a known plate or face is added to a recognition list inside Protect, and once enrolled, future matches generate labeled events instead of generic "vehicle detected" or "person detected" alerts. Unrecognized plates and faces still get logged and are searchable by object type, but they won't carry a name or household label unless someone enrolls them first — enrollment is a setup task, not something the camera figures out on its own.
This matters for a driveway gate on a larger property: a family with several cars, regular staff, and periodic contractors will need someone maintaining that enrollment list, removing departed staff and adding new vehicles, or the system will keep generating unlabeled alerts for people it should already know. That's a process question worth raising with your installer before you rely on recognition for daily convenience rather than just after-the-fact search.
What can a recognized plate or face actually trigger?
Within Protect, a recognized plate or face produces a labeled, searchable event rather than a plain motion alert — that much is part of the same recognition and labeling workflow Ubiquiti describes. Whether that event should also be wired to open a gate, unlock a door, or fire some other automation is a configuration and design question to put to your installer rather than something to assume from the recognition feature alone; ask specifically what your gate hardware and access platform support, and have them show you the automation set up and tested before you rely on it. Treat an automatic unlock on recognition as a trade-off worth discussing deliberately: it buys convenience but also raises the stakes of a misread or a cloned plate reaching the first barrier unchecked. Ask your installer to walk through what happens on a false match, what the fallback entry method is if a plate read fails (weather, dirt, glare, trailer hitches obscuring plates), and whether the gate should still require a secondary factor — an app confirmation, a PIN, a card — for anyone other than habitual, pre-approved vehicles.
Where does the recognition data and footage actually live?
Ubiquiti states that "UniFi Protect leverages edge computing for on-camera AI," and that AI analytics are "processed securely your local infrastructure," enabling instant search and real-time facial-recognition notifications without sending that analysis to a cloud service (UniFi Physical Security - Ubiquiti). Separately, Ubiquiti states that all security footage remains local to the UniFi Console to avoid unnecessary cloud storage for maximum data privacy. Remote viewing is handled as an encrypted connection back to that local console rather than storage of the video itself off-site. This local-first architecture is also why camera recognition pairs cleanly with a broader managed security layer — alarm monitoring and live guard response are separate services layered on top of the recorded and recognized footage, not replacements for it. Cave Group's security upgrade work covers how cameras, access control, and monitoring fit together as a designed system rather than separate purchases, and is a good place to start that conversation with your installer.
What should get tested before you rely on gate recognition day to day?
Plate reads depend on camera angle, distance, lighting, and the plate itself being clean and unobstructed — none of which a product spec sheet can guarantee for your specific driveway. Before signing off on a gate camera install, ask for a live test at the actual approach speed and lighting conditions (including after dark, since IR performance and glare behave differently than daylight), a review of what happens when a plate fails to read, and confirmation of which exact camera model is mounted at that position and whether it is on Ubiquiti's current list of AI/G6-series cameras that support recognition at all. A related piece on door station and gate hardware covers the broader decision between a Crestron-native door station and a UniFi Access gate setup if you're still choosing hardware rather than tuning an installed system.
Cave Group designs and installs UniFi Protect systems as part of its layered security work, specifying camera models and placement against what each camera generation actually supports rather than what a product name implies. Ubiquiti's dome and turret camera line spans several generations with different sensors and feature sets (UniFi Dome & Turret Security Cameras - Ubiquiti), which is exactly why confirming the specific model at each mounting position matters more than relying on a family name.
Frequently asked questions
Does every UniFi camera support license plate and face recognition?
No. Ubiquiti specifies that license plate and face recognition "requires an AI or G6 series camera," so older models, lower-tier models, and non-AI/G6 cameras in the same system will not perform this recognition even if the NVR software is up to date.
Does license plate recognition work by matching a photo of my car?
No — it reads the plate as text and compares that string against an enrolled list, which is why Ubiquiti frames it as being able to "find vehicles by type and color" and filter by recognized plates, rather than matching a vehicle's overall appearance.
Where is the recognition processing and footage stored?
On the local UniFi console or NVR. Ubiquiti states that Protect AI analytics are processed on local infrastructure and that all security footage remains local to the console, with cloud backup available only as an added, optional step.
Can a recognized plate automatically open my gate?
That depends on how your installer configures your specific gate hardware and access platform — it isn't guaranteed by the recognition feature alone. Ask your installer to show you the automation working, walk through fallback entry when a read fails, and discuss whether a secondary factor should still be required for anyone outside a small list of habitual vehicles.
Do I need to do anything for the camera to recognize my family's cars or faces?
Yes — recognition requires enrollment. A plate or face has to be added to a recognition list before the camera will label future matches; unenrolled plates and faces are still detected and searchable but won't carry a name automatically.