[Detector Support]: Coral TPU & GPU

[MrCaspan]

Describe the problem you are having

I have a USB Coral TPU and a Nvidia GTX 1080 with -tensor image loaded . Is it worth having both available for decoding or should I only use the Coral TPU. I have 6 cameras 3x 4K, 2x 2K 1x 1K Just not sure if its wroth having both running or not

Version

0.14.1

Frigate config file

N/A

docker-compose file or Docker CLI command

N/A

Relevant Frigate log output

N/A

Install method

Docker Compose

Object Detector

Coral

Screenshots of the Frigate UI's System metrics pages

N/A

Any other information that may be helpful

No response

[hawkeye217]

What inference times are you seeing?

[MrCaspan]

I was seeing ~10ms with GPU TPU and ~10ms with Coral TPU

But there is 10% gpu memory used but 0 GPU power is used

Not sure if this helps?

[NickM-27]

You seem to be misunderstanding. The coral can do one thing and one thing only and that is run inferences from machine learning models. The coral does not do any decoding or motion detection

[NickM-27]

Frigate also does not support running multiple object detectors at the same time so not sure how you're using a tpu and gpu at the same time.

[MrCaspan]

Sorry Nick I meant I did a tested with both of them, IE not at the same time but one at a time to see if one performed better. I am still supprised on the CPU raise of 13%. on either of them. So I am guessing the motion detection is causing the gain in CPU. Let me test with each. with motion off

NM I beleive detection requires motion to be on.. Ill double check but the CPU did settle down over the last hour

[MrCaspan]

I think I know why CPU is high now. I tested the stream using the API URL and this is what I see.... i think it's a little too sensative

[NickM-27]

Yeah seems like it