Coral M.2 Accelerator with Dual Edge TPU …

It took me a bit to get my VM working in Proxmox with this, but it's blazing object detecting now!

This was a huge pain in the A$$ to set up. I purchased it as an upgrade from my coral USB accelorator. Only to find it out was a nightmare to get up and running in on my raspberry pi. Mainly because Google abandoned the project 3 years ago. Meaning they havnt updated anything. One of those things is the version of python. The libraries you need run on an older version of python and Raspberry Pis OS comes with the latest version. I had to find an older version of debian LINUX that still used it and run a docker container just to set everything up. WAS NOT fun and I essentially was so disappointed, havnt used it much since. You can still use it to increase your inference rate for training your Ai model, but keep running into so many issues it's really not worth thr hassle I woukd not suggest buying one unless support and development has inprocd

Received an defective item. The customer service is great!

Attempted to use this Coral TPU with Blue Iris/Code Project.AI.First unit worked for two weeks before it died. Computer would no longer even power on with the TPU installed. Replacement unit never worked - computer would always hang during boot if it was installed.In case it matters, I used the TPU with an M.2 to PCIe adapter as my motherboard doesn't have a spare M.2 slot. YMMV if you install the TPU direct to the motherboard.

I installed this in my unRAID server using a dual PCI-E adapter. I have been using both Coral units in Frigate for several weeks, and it just works. Was surprisingly easy to setup and use.

As others have pointed out, the "Coral" device itself is great; however you need to be extra careful which M.2 version you get. I had to return these for the M.2 A+E key version as I removed a M.2 WiFi card and wanted to replace it with a M.2 Coral.THIS version of the device will likely NOT work with your motherboard, look for the A+E key version instead. That unit works good on Linux (had to manually build a DKMS module for it to work on Ubuntu 24.04) and Frigate integration is pretty straight forward.Frigate shows an inference speed of 7.5ms and the device runs at 51°C.

I use this with the free/open source Frigate NVR software that is connected to my Home Assistant.I have an HP EliteDesk 705 that I took out the wifi/bluetooth card because I use a hardwired LAN connection and put this Coral in instead. Sadly it only detects one of the TPUs, but that's not the Coral's fault, it's a limitation on my computer.

I got mine from another outlet here on amazon but these are AWESOME. I went from an Amcrest AI NVR with like 32 poe ports (the $600.00 one) which was detecting deer as people and setting my sirens off to an i5, 64gb ram, sapphire nitro Rx 750 and this chip. No more false positives. I had to get an older version of MX Linux that came with python 3.9 then install pycoral and some other stuff but once you make a few efforts it's all downhill. I'm ready to get about 3-4 more and set them up in mesh mode via CPAI.