Cube IDE 2.0.0 and stm32ai-modelzoo-services: Board programs OK, but screen remains black

[emmanuel_]

I encountered a bug with STM32Cube IDE 2.0.0 and \stm32ai-modelzoo-services. When I try to deploy my application using the command:

python .\stm32ai_main.py --config-path . --config-name .\user_deployment_n6_config.yaml

the board programming seems to complete successfully, but the screen remains black.

However, if I use an older version, STM32Cube IDE 1.18.1, the application deployed with the same Python script works perfectly.

tools:
stedgeai:
optimization: balanced
on_cloud: False
path_to_stedgeai: C:/ST/STEdgeAI/3.0/Utilities/windows/stedgeai.exe
path_to_cubeIDE: C:/ST/STM32CubeIDE_1.18.1/STM32CubeIDE/stm32cubeide.exe

# path_to_cubeIDE: C:/ST/STM32CubeIDE_2.0.0/STM32CubeIDE/stm32cubeide.exe

deployment:
c_project_path: ../application_code/image_classification/STM32N6/
IDE: GCC
verbosity: 1
hardware_setup:
serie: STM32N6
board: STM32N6570-DK
stlink_serial_number: "0040003D3234511733353533"

(stm32ai-modelzoo-services4) PS C:\Users\papam\PycharmProjects\stm32ai-modelzoo-services4\image_classification> python .\stm32ai_main.py --config-path . --config-name .\user_deployment_n6_config.yaml
[INFO] : Running `deployment` operation mode
[INFO] : Using provided class names from dataset.class_names
[INFO] : ClearML config check
[WARNING] The usable GPU memory is unlimited.
Please consider setting the 'gpu_memory_limit' attribute in the 'general' section of your configuration file.
[INFO] : The random seed for this simulation is 123
Loading model from ./tf/src/experiments_outputs/2026_02_25_07_34_35/quantized_models/quantized_model.tflite
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
input_shape= (1, 128, 128, 3)
image_size= (128, 128)
[INFO] : Generating C header file for Getting Started...
[INFO] : Please on STM32N6570-DK toggle the boot switches to the left and power cycle the board.
application_code/audio/STM32N6
application_code/face_detection/STM32N6
application_code/hand_posture/STM32F4
application_code/image_classification/STM32N6
[WARNING]: Submodule 'application_code/image_classification/STM32N6' has uncommitted changes. Please commit or stash them.
loading model.. model_path="./tf/src/experiments_outputs/2026_02_25_07_34_35/quantized_models/quantized_model.tflite"
loading conf file.. "../application_code/image_classification/STM32N6/stmaic_STM32N6570-DK.conf" config="None"
"Debug" configuration is used
[INFO] : Selected board : "STM32N6570-DK Getting Started Image Classification (STM32CubeIDE)" (stm32_cube_ide/Debug/stm32n6)
[INFO] : Compiling the model and generating optimized C code + Lib/Inc files: ./tf/src/experiments_outputs/2026_02_25_07_34_35/quantized_models/quantized_model.tflite
setting STM.AI tools.. root_dir="", req_version=""
Cube AI Path: "C:\ST\STEdgeAI\3.0\Utilities\windows\stedgeai.exe".
[INFO] : Offline CubeAI used; Selected tools: 11.0.0 (x-cube-ai pack)
loading conf file.. "../application_code/image_classification/STM32N6/stmaic_STM32N6570-DK.conf" config="None"
"Debug" configuration is used
compiling... "quantized_model_tflite" session
model_path : ./tf/src/experiments_outputs/2026_02_25_07_34_35/quantized_models/quantized_model.tflite
tools : 11.0.0 (x-cube-ai pack)
target : "STM32N6570-DK Getting Started Image Classification (STM32CubeIDE)" (stm32_cube_ide/Debug/stm32n6)
options : --st-neural-art default@../application_code/image_classification/STM32N6/Model/user_neuralart_STM32N6570-DK.json --input-data-type uint8 --inputs-ch-position chlast --output-data-type float32
"series" value is not coherent.. stm32n6 != stm32n6npu
results -> RAM=245,760 IO=0:0 WEIGHTS=419,457 MACC=0 RT_RAM=13 RT_FLASH=185,247 LATENCY=0.000
[INFO] : Optimized C code + Lib/Inc files generation done.
[INFO] : Building the STM32 c-project..
deploying the c-project.. "STM32N6570-DK Getting Started Image Classification (STM32CubeIDE)" (stm32_cube_ide/Debug/stm32n6)
updating.. Debug
-> s:copying file.. "network.c" to ..\application_code\image_classification\STM32N6\Model\STM32N6570-DK\network.c
-> s:copying file.. "network_ecblobs.h" to ..\application_code\image_classification\STM32N6\Model\STM32N6570-DK\network_ecblobs.h
-> u:copying file.. "stai_network.c" to ..\application_code\image_classification\STM32N6\Model\STM32N6570-DK\stai_network.c
-> u:copying file.. "stai_network.h" to ..\application_code\image_classification\STM32N6\Model\STM32N6570-DK\stai_network.h
-> u:copying file.. "network_atonbuf.xSPI2.raw" to ..\application_code\image_classification\STM32N6\Model\STM32N6570-DK\network_atonbuf.xSPI2.raw
-> u:copying file.. "app_config.h" to ..\application_code\image_classification\STM32N6\Application\STM32N6570-DK\Inc\app_config.h
-> updating cproject file "C:\Users\papam\PycharmProjects\stm32ai-modelzoo-services4\application_code\image_classification\STM32N6\Application\STM32N6570-DK\STM32CubeIDE" with "NetworkRuntime1100_CM55_GCC.a"
building.. Debug
flashing.. Debug STM32N6570-DK
using the supplied stlink_serial_number: 0040003D3234511733353533
[INFO] : Deployment complete.
(stm32ai-modelzoo-services4) PS C:\Users\papam\PycharmProjects\stm32ai-modelzoo-services4\image_classification>

[Julian E. · Accepted solution]

Hi @emmanuel_,

The model zoo deployment applications for N6 are nothing more than these standalones getting started.

STM32N6-AI | Software - STMicroelectronics

In the case of Image classification, they indeed specify to use cubeIDE 1.17.0:

GitHub - STMicroelectronics/STM32N6-GettingStarted-ImageClassification: An AI software application package demonstrating simple implementation of image classification use case on STM32N6 product.​ · GitHub

There is maybe something wrong with the version 2.0.0.

The N6 is a very particular target...

Have a good day,

Julian