Table of Contents
Most proactive security features fail for a simple reason: homeowners stop using them. The technology keeps working, but the person who installed it stops paying attention to it.
A camera can trigger notifications for every person detection, generate dozens of alerts a day, and within weeks the homeowner has either disabled it out of annoyance or learned to ignore it. But a disabled feature provides no security value, no matter how sophisticated the underlying detection is.
The False Alarm Problem at Scale
A U.S. Department of Justice Office of Justice Programs guide found that most burglar alarm calls sent to police are false, between 94% and 98%. False burglar alarms can consume a significant share of police workload, accounting for roughly 10% to 25% of all calls to police in some jurisdictions.
The guide addresses police dispatches, not every app alert or consumer-facing alarm notification, but it underscores a deeper problem for home security systems. Police officers are obligated to respond diligently to each call. They can’t simply choose to ignore a call even if they suspect it is a false alarm. But for the average homeowner, regular false alarms lead to lower trust in their security system. Over time, that means homeowners check alerts less and pay less attention to their system notifications.
Parks Associates research sharpens the picture at the homeowner level. The firm reports that 62% of security system owners experienced at least one false alarm in the past 12 months, and that 53% of those owners cited a non-person as the cause. Nearly half of security system owners said their system triggers too many false alarms
If a homeowner gets a notification for every delivery driver, every neighbor walking a dog, and every person passing the property, the alerts become a constant interruption rather than a deterrent. And constant interruptions get switched off or discounted.
Why Simple Person Detection Is Not Enough
Person detection is a solved problem for most modern camera systems. The harder challenge is reading whether a detected person’s behavior represents a genuine threat.
An audio warning that activates with every person detection will fire correctly for a porch pirate but also for the mail carrier, a visiting neighbor, and anyone walking past the camera’s field of view. The system cannot distinguish between them. The homeowner, receiving the same alert for each, loses confidence in the feature.
One camera provider, Vivint, has developed a direct approach to this problem using AI for intelligent triggering. The company’s Smart Deter™ feature, available on the Doorbell Camera Pro and Outdoor Camera Pro 3, and Spotlight Pro is designed to identify behavioral patterns that correlate with a higher probability of criminal intent, such as approaching a package on the porch, dwell time on the property, or approaching a vehicle, and to trigger only on those scenarios. Activating a deterrence response to those patterns rather than on presence alone changes the signal-to-noise ratio of the feature and, in turn, keeps homeowners aware of legitimate threats.
The Engineering Problem
Building a system that distinguishes a delivery driver from a potential burglar or package thief requires more than accurate person detection. It requires a working model of how people actually behave in residential environments: what normal traffic looks like, how legitimate visits differ from suspicious approaches, and what behavioral signals precede a theft rather than follow from innocent activity.
For the human eye, that may seem relatively easy, but for an automated camera, it’s a harder problem than it looks. Approaching a package on the porch, lingering near the door, or moving toward a parked vehicle can each signal a potential threat. Smart Deter handles this on the device itself. Its computer-vision AI evaluates what the camera sees in real time and is built to recognize the behaviors that tend to precede a theft while treating ordinary comings and goings as routine.
Knowing When Detection Works
The security industry has historically measured detection systems on two things: how reliably they catch real threats, and how well they avoid false alarms. Both matter, but neither captures the variable that determines whether a proactive feature actually helps the homeowner day to day: whether it stays quiet during ordinary activity and speaks up for the events that matter, so the homeowner keeps trusting it and leaves it switched on.
A system that is technically capable of detecting a threat but generates enough false positives to get disabled or ignored is, in practice, no more effective than a system with no deterrence capability at all. The metric that counts, then, is detection accuracy paired with a false-positive rate low enough that homeowners keep the feature turned on.
False alarms are a recurring source of frustration for security system owners and often contribute to cancellations. In security system design, annoyance is a primary variable. A deterrence feature is only as good as its restraint, because one that can’t stay quiet during ordinary activity often simply gets switched off or ignored.
What Is Smart Deter™?
Smart Deter™ is a Vivint feature available on the Doorbell Camera Pro, Outdoor Camera Pro 3, and Spotlight Pro that uses on-device computer-vision AI to distinguish normal activity from suspicious behavior. When it identifies a potential threat, it responds with a deterrence behavior matched to the device—a bright LED ring and 90 dB speaker tone on the Doorbell Camera Pro, and a sweeping spotlight, strobing effect, and additional lighting behaviors on the Spotlight Pro and Outdoor Camera Pro 3—rather than triggering on every person detection. On the Doorbell Camera Pro, package detection also notifies the homeowner when a package is delivered, plus it deters potential thieves with light and sound.