@NickM-27 commented on GitHub (Sep 21, 2023):
closing as the PR has been merged
@NickM-27 commented on GitHub (Sep 21, 2023):
this may just be an issue with the new motion detection being more sensitive causing the detector to run harder.
@github-actions[bot] commented on GitHub (Oct 22, 2023):
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.…
@SuperMaximus1984 commented on GitHub (Jan 24, 2024):
I had the same issue with yolov7x-640.trt model, however changing to yolov7-640.trt dropped CPU/GPU usage significantly. Smaller models make…
@kajer commented on GitHub (Mar 12, 2024):
I made the change from .12 to .13 and followed the docs for model updates as well. I even went through the exercise of adding substream detectors to my…
@NickM-27 commented on GitHub (Sep 16, 2023):
interesting, that is not what I expected. @NateMeyer any ideas why this would be the case
@NickM-27 commented on GitHub (Sep 17, 2023):
So one thing that is important to understand is that when you request a large timeframe (more than 2 hours) this logic has to make multiple requests…
@NickM-27 commented on GitHub (Sep 17, 2023):
https://github.com/blakeblackshear/frigate/pull/7846 should fix this
@NateMeyer commented on GitHub (Sep 17, 2023):
I don't know of anything obvious to check. I haven't seen this behavior before.
@robotnikz commented on GitHub (Sep 18, 2023):
OK, thanks for your help guys. I'll try to set up a fresh install of 0.13 then at a later point. Maybe there's just something with my system…
@coledeck commented on GitHub (Sep 17, 2023):
Oh cool! I'll give it a shot if I can figure once I figure out how to use dev builds.
Your theory does make sense, I could do some more testing…
@coledeck commented on GitHub (Sep 17, 2023):
Ok ok, it looks like it is consistent if I pick the same timeframe. Before I was trying different timeframes and getting less and less content.
For…
@robotnikz commented on GitHub (Sep 16, 2023):
Also my inference raised from ~20ms to ~60ms :
The first time I tried it, worked fine. Gave me a 2.1GB recording of a few days of clips.
The next time I tried it, gave me a 7.5MB recording…
@NickM-27 commented on GitHub (Sep 16, 2023):
Please post a screenshot of the system page
@NickM-27 commented on GitHub (Sep 16, 2023):
This is covered in the release notes, downgrading is not supported without restoring a backup of the db or deleting the db and starting fresh.