Object detection on the K210 from m5stack

Question/Issue:
I have a trained object detection model and I want to use it on the K210 to make the preductions and then send those prediction via serial port to a raspberry pi which handles the GUI for the user.
Is there a way to port the edge impulse model to the k210 using maixpy. I’ve already converted the float32-model into a kmodel but the result of classification is a feature map and I don’t really know how to convert this feature

Project ID:
216324

Hi @Merten

Welcome to the forum, we had a previous post on this.

Please see this similar conversation for some helpful steps to try:

Best

Eoin

I’m trying to put a model on a k210 as well but i cant convert the .lite model to a kmodel due to a few different errors.

Could you provide the steps you took so i can follow? Thanks!