If your phone buzzes every time a moth flies past the porch light, you have a PIR problem. If it records every passing car but misses the person who walked up your steps, you have a detection-zone problem. Person detection and PIR solve different halves of this puzzle, and understanding which one your camera uses explains most of those useless alerts.
PIR — passive infrared — senses moving warmth. AI person detection analyses the shape in the frame and decides whether it looks like a human. AI is usually better at cutting junk alerts in busy scenes, while PIR is simpler, cheaper to run, and kinder to batteries. The right pick depends on where the camera sits and what you need it to ignore.
How a PIR sensor actually decides to record
A PIR sensor does not see an image at all. It watches for changes in infrared energy — essentially, a warm shape moving across its field of view against a cooler background. When a person walks past, the sensor registers that heat contrast and wakes the camera to record or send an alert.
This simplicity is both the strength and the weakness. Anything warm that moves can trip it: a large dog, a car engine pulling into a driveway, warm air from a dryer vent, even sun shifting across a wall. Small animals close to the sensor look large, so a cat at close range can read like a person further off. PIR also struggles when the background itself is hot — on a warm afternoon a person is barely warmer than sun-baked pavement, so the contrast the sensor needs fades.
What person detection means inside a modern camera
Person detection starts where PIR stops. Instead of asking did something warm move, the camera asks does that shape look like a person. A small processor on the camera — or the manufacturer's cloud service — compares moving shapes against patterns learned from thousands of human silhouettes: head, shoulders, torso, legs, the way limbs move together.
When the match is strong enough, the clip gets tagged as a person and you get that specific alert. Some cameras let you record everything but only notify for people, which is the setup most households actually want.
The catch is that this analysis needs a usable image. If the person is a small silhouette at the far edge of the frame, backlit into a dark blob, or half-hidden behind a fence, the algorithm has little to work with and may stay silent.
Where PIR still wins: doorways, sheds, and battery cameras
PIR remains the better tool in a few specific spots. A camera watching a narrow hallway, a shed door, or a garage interior has a simple job: anything warm moving there is worth recording. There is little passing traffic to filter, so the AI layer adds little while consuming extra power.
Battery life is the other decisive factor. A PIR sensor sips power while it waits, waking the camera only when something moves. Person detection that runs continuously on the video feed draws far more energy. That is why many battery cameras use PIR as the wake-up trigger and run person analysis only on the resulting clip.