Hi, I assume in for HaOS, Frigate and other add-ons only object detection is used. Looking at Google Coral USB, I see it is probably abandoned by Google. Is there a better hardware to use now?
I want to use it with Frigate or any other uses it may have on HaOS.
Also, are there different versions of the Google Coral USB or just one?
Thanks.
I could never get the Coral USB detected by the software even though it showed up in the hardware.
I am not using the same software you mentioned though. I am using Blue Iris with CodeProject.AI.
Just a heads up.
Only one usb
Other PCI and M2 models
Thanks.
Is there a preference of one over the other? I have read the Frigate wiki and doesn’t seem to give an indication. Is there a difference in the smartness of object detection of things? For speed, I can bundle up more and more.
Also, do any all of these allow for learning? i.e. show it shoes and expect it to detect in future.
Depends on your priorities (ex: power consumption vs performance) and your current hardware
That comes down to the models that can be run, which the docs explain
You can train any model, but it’s a very involved process
No. There is no preference. Use what’s available and what works best/easiest for you. You can add multiple usb or other models. Crzynic is correct however.
PCI and M2 offer models with dual TPU.
Dual TPU PCI likely require special PCI adapter to use both TPU. I lost Weblink.
Thanks.
Is there anything else on HaOS that Google coral can be used for other than Frigate image processing?
There are a few other things, but you should keep in mind that a coral can only be accessed by one process at a time, so it can’t be used for multiple services at the same time. That is one of the advantages of using a iGPU or GPU, it can be used for many different features / apps at the same time.
Thanks.
Is there a documented list of what else can coral or similar detectors be used for other than with Frigate?
Google(search for) in web browser “coral tpu uses”
