bizier

I searched a bit and see what you mean: cool demos but nothing available for sale to consumers. And I bet the enterprise things cost a lot more than installing a handful of PIR sensors. So until someone makes awesome open source software for this, analytics is not worthwhile. So the choices are: carefully tweak the motion settings as you did to get the false positive rate down a little, or totally nail the problem by using PIR sensors. PIR sensors would mean switching NVRs, as the Lorex LNR280 doesn't seem to support triggering/tagging the video stream based on external inputs. If I go the PIR route: I'm still wondering, do I have to place it some distance away from the (IR-emitting) camera to be effective? Thanks again!

Thanks for the advice! What is it about the PIR sensors that make them better at minimizing false positives? Is it that they're not right next to the IR emitter? (I imagine the IR might attract the bugs; the direct reflection might also not help matters.) If so, I imagine I'd have to place the PIR sensors some distance away from the camera to be effective - how far? I wonder also if it'd be possible to get a similar effect by disabling the camera's IR emitters and use some external ones instead. I would think that you could do a lot better in software at discriminating person-like motion from bug-like motion, but if such software isn't available off-the-shelf, I don't really have time to invest in advancing the state of the art myself.

I got the Lorex 8-camera LNR280 NVR as part of the Costco package with 4 cameras. Seems to work decently well: good capacity (2 TB on the included drive) the web+plugin, Android tablet, and Android phone interfaces seem okay (but not perfect) the builtin PoE switch is handy ...except the motion detection has way too many false positives. There were over 100 events last night, I think from bugs. A few so far this morning also. Has anyone been able to do better? There are two controls: draw an area and slide the sensitivity. Seems like it detects nothing with no area selected, so the red grid must be where you want to detect motion. And seems like it detects nothing with the slider all the way to the left, so I have it one notch to the right. I'll repeat those experiments later to be sure. My yard has a lot of plants and trees moving in the wind, small creatures (bugs, birds, squirrels, and neighborhood cats) going by, and some areas where you can see neighbors' yards or the streets. I've tried to select only non-plant parts of my own yard, but the motion detection is still not that useful. Should I be expecting this NVR to handle this? If not, would some other one be better? I'm new to this.