Everything moves. Almost none of it matters.

A conventional perimeter system answers one narrow question: did enough pixels change inside this box? A tree line in the wind, rain on the lens, a fox crossing the yard, headlights sweeping a fence, a delivery driver taking a shortcut he has taken every week for a year — all of these change pixels, all of them cross a line, and each produces an alarm carrying exactly the same weight as a person going over the fence at three in the morning.

The consequence is not a technical annoyance; it is an operational failure. Guards learn, correctly, that almost every alert is nothing, and they stop looking. Sites respond by narrowing detection zones, shortening armed hours and raising thresholds until the system is finally quiet — which is to say, until it is also blind. Meanwhile the intrusion that mattered arrives in a long queue of alerts that didn't matter, and on screen it looks exactly like all of them.

Vality reads the boundary differently. It doesn't ask whether something moved; it asks what the movement means. A person walking a fence line parallel to it, at working pace, in daylight, during a scheduled contractor window, is not the same event as a person approaching that same fence perpendicular, slowing, stopping twice and standing in the camera's blind corner for four minutes after close. To a detector the two are near-identical bounding boxes. As behavior they have nothing in common.

That distinction is only available to a system that reads sequences rather than frames, and that knows what this zone is for, who is entitled to be in it, and what it normally looks like at this hour. What comes out is not a longer list of detections. It is a shorter list of things worth walking towards.