Guides · Home Security
Mounting a camera inside, looking out through a window, is tempting. It is protected from weather, easy to power and simple to move. It is also one of the most common sources of false alerts, because glass turns the camera into a mirror at exactly the wrong moments.
At night, an indoor lamp or a screen lights the room and the window behaves like a mirror. The camera sees a reflection of its own room and anything that moves in it, such as a ceiling fan, a pet, or a person walking past. Motion in the reflection looks like motion outside.
Infrared night vision adds a second problem. Most cameras illuminate the scene with built-in infrared LEDs. When the lens sits close to the glass, that light bounces straight back as a bright halo or white wash, which hides the scene and can flicker as the exposure keeps adjusting. Rain, condensation and dust on the glass scatter the light further.
Finally, headlights and streetlights sweep across a window and produce rapid brightness changes. A motion detector that only measures pixel change reads this as movement, while a system that classifies what it sees can recognize that no person or vehicle is actually present.
Software can only compensate so much, so start with the physical setup. If you must shoot through a window, sit the lens flush against the glass, or use a rubber or foam collar around it so no room light leaks in between the lens and the pane. Switch off or cover lights and screens that sit behind the camera, and avoid placing it where a lamp lines up with the window.
Where possible, turn off the camera’s built-in infrared light and rely on an external light or exterior lighting instead. Alternatively, choose a weatherproof outdoor camera so no glass sits in the way. For entry points such as a front door, an outdoor position usually gives a cleaner result and a better view of faces and approach paths.
Lowering sensitivity across the board is a blunt fix that can cause the camera to miss real events. A better approach is to use the tools that target the actual problem. Activity zones let you exclude the part of the frame that is mostly reflection, such as the glass of a neighboring window or the area lit by a lamp. Object filters that alert only for people, vehicles or packages ignore moving light and shadow.
This is where behavioral AI helps. Rather than reacting to any change in brightness, the system looks at what is moving and how: whether the object has the shape and motion of a person, whether it is approaching a door, whether it lingers. Reflections tend to mirror movements inside the room and rarely behave like someone approaching from outside, which gives a classifier more to work with than raw pixel change.
After any change, review a few nights of footage. Look at which clips triggered, what time they happened, and whether the same lamp, fan or passing car appears each time. Adjust one setting at a time so you can see what made the difference. If the same cause keeps producing alerts, fix the cause, whether that is moving a lamp or masking an area, instead of loosening the whole system.
If you still see halos, ghost images or constant triggers after sealing the lens and adjusting zones, glass is probably the wrong place for that camera. Moving it outdoors, under an eave, is a small change that usually removes the problem entirely, and it also improves the quality of the evidence you capture.
Reducing false alerts is about making each notification worth acting on. A camera that rarely cries wolf is one you will trust and keep switched on. Sentrick Shield focuses on property protection through behavioral AI that distinguishes real activity from noise, and is designed to work alongside good placement rather than replace it.