News:

Rule #1 - Be Patient - Rule #2 - Don't ask when, if you don't contribute - Rule #3 - You have coding skills - LinuxMCE's small brother is available: http://www.agocontrol.com

Main Menu

Motion Detection Cameras for Presence Detection?

Started by Matthew, January 15, 2008, 12:19:53 AM

Previous topic - Next topic

Matthew

Is LMCE's existing support for motion detection cameras, used for alarm systems, a suitable foundation for supporting presence detection? So the cameras would always watch the rooms, and detect in which rooms there were people, and activate MDs, lights etc. Maybe even watch people leave, and switch from media/automation service into alarm mode until deactivated. Can it be calibrated to ignore moving pets, and just track people's presence in different rooms?

If it were really good, it could register users and track them individually, maybe by facial recognition. Or maybe people could wear watches or other jewelry/accessories that let LMCE distinguish among them. That's pretty far out, but even just telling whether there's >0 people in a room for more than just an alarm would be valuable.

colinjones

Would be very nice, but I suspect that building a state table/engine from the vision/motion information would produce very unreliable results in terms of presence. Even more so than IR detection. You need to be able to reliably determine state changes from occupied to unoccupied and back again, otherwise the whole house would start doing weird stuff! Even with very careful camera placement, I think that is going to be difficult to detect. And you certainly couldn't rely on pure motion alone, that would be the same as IR - you'd have to keep waving your hands around so that LMCE doesn't turn off the room :)