Announcing Official Support for Seeed’s Grove Vision AI Module

Edge Impulse and Seeed have partnered together to support Seeed's Grove - Vision AI Module with Edge Impulse, enabling users to now acquire, develop, and deploy vision-based ML applications with Edge Impulse Studio and Edge Impulse SDK.

This is a companion discussion topic for the original entry at

nice! does it run fomo? and if so what sort of performance can one expect?

The screenshots shown are from a model running a FOMO model with an overall inference time of 0.090 seconds.

1 Like

ok, thanks!
Another question. Can I run the board standalone and program it with arduino code? Or do I need something like the seeed xiao microcontroller board to connect to the AI vision board?
I would need gpio pins control for most of my projects, to control relays, switches and so on…


Seeed’s Grove - Vision AI Module is compatible with the XIAO ecosystem and Arduino. So design, build, train, test and deploy your model from Edge Impulse Studio using either the Edge Impulse Arduino library or the Edge Impulse C++ library that is custom built for your model.

Thanks. I downloaded my model as an arduino library, but there is no example how to use it for the grove ai vision board as there is for other boards like esp32, nicla vision, nano sense 33 ble, etc, (and I could not find any info how to use grove ai vision with xiao board+arduino either), please keep in mind im not an expert, just a curious tech interested person trying to learn (it can be a lot of info to take in sometimes)

Can someone point me in the right direction?


Hi @Instant_exit,

Is this what you are looking for? Seeed Grove - Vision AI Module - Edge Impulse Documentation

Thanks! I might have missed it, but if you deploy an arduino library, there are included examples for other boards, but I could not find it for the grove ai vision.

(It is also confusing how you use the board connected to a microcontroller in the xiao series, but that is not really your fault, I tried looking at the seeed documentations. )


Hi @Instant_exit,

You are correct…at this time, a “supported board” means we have a pre-compiled firmware for it that will communicate with the Edge Impulse Studio (usually via CLI or WebUSB). It does not necessarily mean we have a full Arduino library for it (yet).

Hello, The EI instructions ask for installing the latest Bouffalo Lab Dev Cube software, but only a .exe (Windows version) is really available. When we download de sw for all platforms:

The version for MacOs has no working version for instalation:
(the “bflb_iot_tool-macos” file is not executable.