An alert system that cries wolf gets ignored, and outdoors is where that happens fastest. A fox crossing the garden at 2am, a cat on the fence, or a bin lid lifting in the wind will each generate a notification indistinguishable in urgency from an actual intruder — right up until the household starts dismissing all of them without looking.
The fixes are mostly about geometry and sensor choice, and most of them cost nothing beyond the time to apply them.
Why a basic sensor sees a cat as a person
A passive infrared sensor detects a change in infrared radiation across its field of view — in plain terms, something warm moving. It has no concept of shape, size or species. What it actually measures is closer to how much of the sensor's view changed, and how fast, and a small warm animal close to the sensor can produce a change of very similar magnitude to a person further away.
This is not a fault in cheap sensors specifically. It is the nature of the technology, and it is why outdoor PIR sensors have historically had a reputation for false alarms that indoor ones mostly avoid — indoors, cats and foxes are rarer visitors.
Detection zones and exclusion masks
Better sensors and most outdoor cameras let you define a detection zone — the part of the frame that actually triggers an alert — separately from the full field of view. A driveway camera does not need to alert on movement in the sky or the neighbour's hedge, only on the path and gate.
The specific, high-value use of this is an exclusion mask over the bottom metre of frame, which is where most small animals actually appear relative to the camera, while people still register clearly higher in the frame. This single adjustment removes a large share of animal triggers without reducing genuine sensitivity at all.
Where the hardware only offers a single global sensitivity slider rather than zones, the next best tool is mounting height, covered next.
Mounting height is free tuning
A sensor or camera mounted low, at fence height, sits directly in the path of every cat, fox and hedgehog that crosses the garden. The same device mounted higher and angled down sees people clearly while most small animals pass below its effective detection zone or appear too small in frame to register as a person-sized object.
This is the same underlying principle as camera placement generally — see our piece on where to put an outdoor camera — but it matters specifically for false-alarm reduction: height that improves coverage also, as a side effect, discriminates against small animals.
Camera-based detection filters far better than PIR alone
Modern outdoor cameras increasingly include on-device person and vehicle detection, which analyses the shape of what moved rather than just the fact that something did. This is a meaningfully different technology to a basic PIR sensor and, where available, is worth enabling and relying on ahead of the raw motion trigger.
It is not perfect — a person crouched, partially obscured, or captured only side-on can still be missed or misclassified — but it removes the large majority of animal and foliage triggers that a raw PIR cannot distinguish at all. Where budget allows one camera upgrade for false-alarm reduction specifically, this is usually it.
Wind, foliage and things that move without being alive
Not every false trigger is an animal. A washing line, a swinging gate, a bin lid, or a branch moving in wind all produce genuine motion that a basic sensor cannot distinguish from anything else.
Anything that swings in normal wind should be kept out of a detection zone if at all possible — repositioning a line or securing a gate latch is often simpler than trying to tune sensitivity around it.
A minimum trigger duration — requiring movement for slightly longer than an instant flicker — filters out the quick, small motions typical of leaves and shadows while still catching a person walking through, since a person takes a meaningfully longer time to cross a frame than a leaf takes to flutter.
Two-sensor confirmation is the most reliable general-purpose fix: requiring two independent sensors, or a sensor plus a camera classification, to agree before escalating to a loud alert, while still logging single triggers quietly. This is worth building as an automation regardless of which specific hardware caused the original problem.
“Pet-immune” ratings, and their real limits
Many outdoor PIR sensors and floodlights are marketed as pet-immune up to a stated weight, commonly somewhere around 18 to 25 kg. This is a genuine, useful specification, not marketing fluff — it works by weighting the sensor's response toward the lower part of its field of view less heavily, on the assumption that a genuine intruder is tall and a pet is not.
The limit worth understanding is that the rating assumes a typical distance. A cat close to the sensor still occupies a similar apparent size in the detection field as a person further away, which is the same underlying geometry problem covered earlier — so a pet-immune sensor mounted too low, or with an animal that regularly comes very close to it, will still trigger more than the rating suggests. The rating is a real improvement, not a guarantee, and mounting height still does most of the practical work alongside it.
Seasonal recalibration
A garden in March and the same garden in August are different environments for a sensor. Foliage grows into what was a clear detection zone, washing lines go up for summer, and evening light angles change enough to shift where shadows fall.
Treat detection zones and sensitivity as something to check each season, not set once. A five-minute walk-around after a season change, watching what triggers a test alert, catches most of the drift before it becomes an ignored stream of notifications.
A tuning checklist
- Enable person or vehicle detection on any camera that supports it, ahead of raw motion.
- Set a detection zone or exclusion mask over the lowest part of the frame where small animals appear.
- Raise and angle sensors where mounting allows, using the same geometry that improves general coverage.
- Remove or secure anything that swings in wind from inside the active zone.
- Add a minimum trigger duration to filter brief flickers.
- Use two-sensor confirmation for anything that should escalate to a loud alert.
- Re-check after each season, not just at installation.
Done together, these turn a system that alerts on every fox into one the household actually trusts — which is the only version of outdoor detection that gets acted on.