@NickM-27 commented on GitHub (Sep 15, 2023):
Your docker run volume mappings are wrong, please read the installation docs https://docs.frigate.video/frigate/installation#docker
@NickM-27 commented on GitHub (Sep 15, 2023):
your hwaccel_args and input_args are wrong and you should really not be using manual args, should be using ffmpeg presets
you also have not…
@NickM-27 commented on GitHub (Sep 15, 2023):
this is the wrong repo for this request, see https://github.com/blakeblackshear/frigate-hass-addons
@github-actions[bot] commented on GitHub (Oct 16, 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.…
@NateMeyer commented on GitHub (Sep 17, 2023):
@vajonam Can you try just adding an env var to your docker-compose.yml and having it rebuild the model? The value is the index of the GPU as seen…
@vajonam commented on GitHub (Sep 17, 2023):
#Yes. This works. But since my GPU 1 doesn't support encode/decode all the streams fail to load up. But I am use this setting to build the right…
@NateMeyer commented on GitHub (Sep 17, 2023):
Ah, ok. I didn't know if this would affect the encode/decode access or just cuda. I'll add an intermediate variable to apply only to the…
@vajonam commented on GitHub (Sep 17, 2023):
I think it’s specific to my environment as I have a card that only supports cuda and no encode / decode. This solution works for me. Just documention…
@NateMeyer commented on GitHub (Sep 17, 2023):
If I'm understanding correctly, the script change is needed to make it so you wouldn't need to boot twice, first generating the models, then…
@vajonam commented on GitHub (Sep 18, 2023):
Yes. That is the correct that’s how it behaves now. 2 boots one with var CUDA_VISIBLE_DEVICES and one without.
If you need it come up the first…
@NateMeyer commented on GitHub (Sep 18, 2023):
Shouldn't need to even change the python script, just bash. I plan to set/unset the CUDA_VISIBLE_DEVICES just for the conversion process.
@vajonam commented on GitHub (Sep 21, 2023):
Works well tested with latest dev snapshot! thanks!
@NickM-27 commented on GitHub (Sep 15, 2023):
Hey, that's a great point, I hadn't considered that issue before. For now, running once with just the detection GPU to build the models should be a…
@vajonam commented on GitHub (Sep 15, 2023):
well I think I got mine working, deleting the models, restricting frigate container to only the GPU I want for object detection, (encode/decode will…