Running My model on my card

I can’t run my model on my card because of the memory size ( memroy card 1MB and my model has 29MB), is there any efficient solution ?

Hi @Ghazi,

Which board and type of model are you using in your project?

Aurelien

I’am using arduino nano 33 ble and c++ model with “zephyr”

I suppose it is an object detection use case given the size of your model.
The Arduino has limited memory but you can use our FOMO feature. You’ll most likely need to set an image size of 48x48px and select the smaller FOMO model (FOMO MobileNet v2 0.1).

Aurelien

Is there something like that for a project were I’am using an accelerometer sensor?

Accelerometer models are usually very small, tens of KB at most so a bit surprised you can get a 29MB model. Could you share your project ID?

Aurelien

Yes of course, this is my project ID 79673

The model is actually quite small:

Our official support for the BLE Sense is for the Arduino environment, maybe this is related to your Zephyr project that has extra instructions?

Aurelien

Yes, this is some folders and files that i should add to my model to run it on zephyr
edgeimpulse/example-standalone-inferencing-zephyr (github.com)

Have you an idea about how to minimize the size of zephyr project?