"Required Ram or flash size for network is bigger than available Ram or Flash" alert with X-Cube_AI extension

[LForm.1]

Hi, I'm trying to load my Keras model (built for CamVid dataset) into STM tool, but when I try to analyze even with compression set to 8 it gives me the "Required Ram or flash size for network is bigger than available Ram or Flash" alert.

I tried to modify my model also trying to eliminate various layers, but nothing.

the model is:

and this is what I have after the analyze:

any idea? thank you!!

[Houssem CHAABANI]

Dear @LForm.1​ ,

Could you share your .ioc file ?

Kind regards,

Houssem

[LForm.1]

Hi! thank you for your reply.

This is the .ioc file

[LForm.1]

Hi! thank you for your reply.

This is the .ioc file

[Houssem CHAABANI]

Dear @LForm.1​ ,

I tried with STM32CubeMX6.1.0 and with HAR-CNN-Keras module and it's working.

Could you please share your model file.

Kind regards,

Houssem

[LForm.1]

I don't know what is the module file, how can I get it? Sorry I'm new to programming.

[Houssem CHAABANI · Accepted solution]

Dear @LForm.1​ ,

The RAM size in STM32f441 is only 128kbytes and your project need more size of RAM.

It's recommended to use another STM32 microcontroller with larger RAM memory.

Kind regards,

Houssem

[Imen.D]

Hello @LForm.1​ ,

Is your problem resolved ?

If yes, please share the solution and please mark the answer as best by clicking on the "Select as Best" button if it fully solved your issue.

This will help other users find this solution more quickly.

Imen