Hardware for running Custom Vision on Azure IoT Edge?

Question

I have been looking at running Custom Vision at IoT Edge. I would like to do something similar to as what they have done in this demo https://www.youtube.com/watch?v=_K5fqGLO8us. They use a Raspberry Pi 3B+ and it works, but it is quite slow. It takes the device around 3-5 seconds to notice a change.

I would like to do something similar, but I would need a latency a and frame rate about 2-3 FPS (i.e. I would need something 10x better than the Raspberry 3B+). Are there any suitable single-board computers available for my needs? Anyone that has done something similar and can give a recommendation?

score 0 · Answer 1 · answered Mar 08 '23 at 20:37

0

You will need a device with a dedicated GPU to run these kinds of workloads quickly. The Jetson Nano might be an excellent place to start. There are other devices in the Jetson Family that could be a better match.

answered Mar 08 '23 at 20:37

Matthijs van der Veer

3,865
2
12
22

Thanks for the input! I have been looking at some Coral products., these in particular https://coral.ai/products/accelerator. https://coral.ai/products/dev-board. They seem quite promising? – Rossario Mar 09 '23 at 14:11
Any purpose-built device will outperform the Raspberry. – Matthijs van der Veer Mar 10 '23 at 12:41

Hardware for running Custom Vision on Azure IoT Edge?

1 Answers1