If you havn’t already, you can add object and face recogniction to your cameras and also implement notifications to your phone for various events, such as sending a picture when a person is detected outside your house when you are not home etc. I’m using Deepstack for object and face recognition. Face and person detection with Deepstack - local and free!
@sd_dracula
In industry we used to maintain lists of stuff that consumed resources (mainly people’s time) and every month tackle the top 5 on that list.
That way you are saving time to spend on ‘other stuff’
There’s two things that I haven’t worked out about doors on science fiction programs.
You walk up to a door wanting to go through it and it opens, you walk very close to the door because you are doing something else and it doesn’t open - how does it know the difference ?
You fire a weapon (vary depending on the programme) at the door mechanism and the door is either fused shut or it springs open - it’s always a wonder to me that the result always moves the plot along in a manner advantageous to the main protagonists ?
Some very cool ideas in here, thanks
Not sure if the pi4 is strong enough to run constant face detection on 4 different feeds.
Washing machine/dryer is more or less next in line.
While obviously awesome, Star trek doors are pretty far down my list at this time
Liking the calendar TTS, which integration are you using for that? Sick of all the wall tablets default google calendar ding
I would like a way for the ring screen to pop up on top on the wall tablets when someone rings the door bell and auto switch to it.
ROTFLOL, I know I’m also picky about sci-fi gaffage. I’m pretty sure, at least in STNG, the enterprise was constantly scanning neural signals so doors knew what your intention was. I had similar thoughts when I posted above… wait a minute, most doors in houses don’t have that long entry like you have at a shopping mall, where a pir sensor is enough to work well. That would suck, walking past my front door naked in the morning, and have it open as my mail lady is doing her thing, LOL! …or not so funny, getting home from work and finding my dog trapped behind an open door.
Never give up… video tracking would probably work… or some strategically placed proximities (90* beams near doorways, you can have a velocity vector to use). It’s not neural scanning, but it could come close enough.
No, it takes up a lot of memory and CPU so you need either a NUC or a Google Coral Accelerator to get 4 feeds working. I have 8 feeds working on a NUC w. 16GB RAM without any noticable CPU increase. USB Accelerator | Coral. I also have the Coral stick and I am using it for some Machine Learning. It’s 20-30 times faster than the NUC by the way.
