- Per camera/device event recording delay parameter.
- Make Person events standalone events not motion subsets. (All I want is person events. I don’t care about seeing my car drive into my garage or my garage roomba doing its job.)
I got this to monitor Amazon Key deliveries. (Amazon’s deal with liftmaster/chamberlain to allow in garage deliveries.) Without jumping through hoops it doesn’t work well at all for that purpose. The hardware/picture quality is fine for my needs. The basic software parts are there too but, unless I just can’t figure it out, are a bit brain dead.
Here is my process and the issues. I first put it in the garage facing the garage door. The garage door, ironically enough, takes 12 seconds to rise. So by the time someone walks in the cloud recorded event is over. (Yes I’ve got a sd card and can manually go see if that particular event was my Amazon dude but what a needless waste of time.) I then positioned it near the garage door so it didn’t ‘see’ the garage door opening figuring the person walking in would trigger the motion. That didn’t work either because it quickly senses the change of light as the door opens.
SO… what is needed is a per device event recording delay time in its settings menu. All this would do is when an event is triggered is to wait the event recording delay time before it recorded the 12 seconds and sent it off to (rumored) China. I’m not asking for more of your cloud time. 12 seconds is fine if you get the right 12 seconds. You have timers already for cool down period ect. So this addition would be super easy, barely an inconvenience.
My second gripe is your so called Person detection implementation. As far as I can figure it is a subset of general motion detection but not a trigger itself. In my use as described I wouldn’t need a event recording delay time if the event started with the actual detection of a person. But what seems to happen is a motion event is triggered and then the software looks to see if a person shows up within the 12 second recording window. (Again digging through ones sd card looking for people is not an acceptable solution in my opinion.) Based on your person detection video you claim this, so called, AI is done on the camera itself and not after the fact on the cloud recorded event. So if I’m looking for person triggered events then the person could be the trigger rather than random motion prior, as nothing needs to be uploaded to the cloud for analysis if it’s truly done on camera.
Anyway, in my opinion, not quite there yet but the bones are good so I’m looking forward to software based improvements.
ps. (I don’t actually care if China is storing videos of my garage. It’s not that interesting of a garage.)