To create an image classifier running on the STM32H747I-DISCO board with Teachable Machine

[FDREV.1]

Hi ST member,

I would like to create an image classification application on STM32 with Teachable Machine like https://wiki.st.com/stm32mcu/wiki/AI:How_to_use_Teachable_Machine_to_create_an_image_classification_application_on_STM32

In first time, I create a model with these parameters :After, I select Tensorflow Lite, check Quantized and export my model.

I try different parameters.

After, I want to add my model in CubeMX (like https://www.youtube.com/watch?v=NDshmSH7WnA&list=PLnMKNibPkDnG9IC5Nl9vJg1CKMAO1kODW&t=1s).

I select these parameters (X-Cube-AI version 6.0.0) :Then, I analyze my model. The analysis returns an error. The model is too big :And if I integrate network.c, network_data.c, network.h and network_data.h into FP-AI-VISION1, this project doesn't compile :

c:\st\stm32cubeide_1.6.0\stm32cubeide\plugins\com.st.stm32cube.ide.mcu.externaltools.gnu-tools-for-stm32.9-2020-q2-update.win32_1.5.0.202011040924\tools\arm-none-eabi\bin\ld.exe: STM32H747I_DISCO_PersonDetect_Google_CM7.elf section `.axiram_section' will not fit in region `AXIRAM'

c:\st\stm32cubeide_1.6.0\stm32cubeide\plugins\com.st.stm32cube.ide.mcu.externaltools.gnu-tools-for-stm32.9-2020-q2-update.win32_1.5.0.202011040924\tools\arm-none-eabi\bin\ld.exe: region `AXIRAM' overflowed by 240096 bytes

I think my parameters of my model is wrong.

[GK.31]

Is there a solution？