Deploy an image classification model in STM32N6570-DK discovery board

[diama13]

I follow the instructions found in  https://github.com/STMicroelectronics/stm32ai-modelzoo-services/blob/main/image_classification/deployment/README.md in order to deploy an image classification model in STM32N6570-DK board. 

When i launch (in ubuntu) 

python stm32ai_main.py --config-path ./config_file_examples/ --config-name deployment_n6_mobilenet_v2_config.yaml

i get the following error 

[WARNING] The usable GPU memory is unlimited.
Please consider setting the 'gpu_memory_limit' attribute in the 'general' section of your configuration file.
[INFO] : Running `deployment` operation mode
[INFO] : ClearML config check
[INFO] : The random seed for this simulation is 123
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
[INFO] : Generating C header file for Getting Started...
loading model.. model_path="../../../stm32ai-modelzoo/image_classification/mobilenetv2/ST_pretrainedmodel_public_dataset/flowers/mobilenet_v2_0.35_128_fft/mobilenet_v2_0.35_128_fft_int8.tflite"
loading conf file.. "../../application_code/image_classification/STM32N6/stmaic_STM32N6570-DK.conf" config="None"
"n6 release" configuration is used
[INFO] : Selected board : "STM32N6570-DK Getting Started Image Classification (STM32CubeIDE)" (stm32_cube_ide/n6 release/stm32n6)
[INFO] : Compiling the model and generating optimized C code + Lib/Inc files: ../../../stm32ai-modelzoo/image_classification/mobilenetv2/ST_pretrainedmodel_public_dataset/flowers/mobilenet_v2_0.35_128_fft/mobilenet_v2_0.35_128_fft_int8.tflite
setting STM.AI tools.. root_dir="", req_version=""
Cube AI Path: "/opt/st/STEdgeAI/2.0/Utilities/linux/stedgeai".
[INFO] : Offline CubeAI used; Selected tools: 10.0.0 (x-cube-ai pack)
loading conf file.. "../../application_code/image_classification/STM32N6/stmaic_STM32N6570-DK.conf" config="None"
"n6 release" configuration is used
compiling... "mobilenet_v2_0_35_128_fft_int8_tflite" session
model_path : ['../../../stm32ai-modelzoo/image_classification/mobilenetv2/ST_pretrainedmodel_public_dataset/flowers/mobilenet_v2_0.35_128_fft/mobilenet_v2_0.35_128_fft_int8.tflite']
tools : 10.0.0 (x-cube-ai pack)
target : "STM32N6570-DK Getting Started Image Classification (STM32CubeIDE)" (stm32_cube_ide/n6 release/stm32n6)
options : --st-neural-art default@../../application_code/image_classification/STM32N6/Model/user_neuralart.json --input-data-type uint8 --inputs-ch-position chlast --output-data-type float32
[returned code = 255 - FAILED]
$ cwd: None
$ args: /opt/st/STEdgeAI/2.0/Utilities/linux/stedgeai generate --target stm32n6 -m ../../../stm32ai-modelzoo/image_classification/mobilenetv2/ST_pretrainedmodel_public_dataset/flowers/mobilenet_v2_0.35_128_fft/mobilenet_v2_0.35_128_fft_int8.tflite --output /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46 --workspace /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46 --st-neural-art default@../../application_code/image_classification/STM32N6/Model/user_neuralart.json --input-data-type uint8 --inputs-ch-position chlast --output-data-type float32
ST Edge AI Core v2.0.0-20049

PASS: 0%| | 0/82 [00:00<?, ?it/s]
PASS: 9%|▊ | 7/82 [00:00<00:01, 39.76it/s]
PASS: 13%|█▎ | 11/82 [00:00<00:02, 34.83it/s]
PASS: 24%|██▍ | 20/82 [00:00<00:01, 54.28it/s]
PASS: 32%|███▏ | 26/82 [00:00<00:01, 47.81it/s]
PASS: 39%|███▉ | 32/82 [00:00<00:01, 44.83it/s]
PASS: 45%|████▌ | 37/82 [00:00<00:01, 43.41it/s]
PASS: 51%|█████ | 42/82 [00:00<00:00, 44.66it/s]
PASS: 57%|█████▋ | 47/82 [00:01<00:00, 46.05it/s]
PASS: 65%|██████▍ | 53/82 [00:01<00:00, 46.39it/s]
PASS: 71%|███████ | 58/82 [00:01<00:00, 43.25it/s]
PASS: 77%|███████▋ | 63/82 [00:01<00:00, 39.67it/s]
PASS: 83%|████████▎ | 68/82 [00:01<00:00, 41.47it/s]
PASS: 89%|████████▉ | 73/82 [00:01<00:00, 40.27it/s]

>>>> EXECUTING NEURAL ART COMPILER

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

/opt/st/STEdgeAI/2.0/Utilities/linux/atonn -i "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0.onnx" --json-quant-file "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0_Q.json" -g "network.c" --load-mdesc "/opt/st/STEdgeAI/2.0/Utilities/configs/stm32n6.mdesc" --load-mpool "/home/test/Work/ST/stm32ai-modelzoo-services/application_code/image_classification/STM32N6/Model/my_mpools/stm32n6-app2.mpool" --save-mpool-file "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool" --out-dir-prefix "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/" --all-buffers-info --no-hw-sw-parallelism --cache-maintenance --enable-virtual-mem-pools --native-float --optimization 3 --Os --Omax-ca-pipe 4 --Ocache-opt --enable-epoch-controller --output-info-file "c_info.json"

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

>>> Shell execution has FAILED (returned code = 1)

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

$ /opt/st/STEdgeAI/2.0/Utilities/linux/atonn -i /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0.onnx --json-quant-file /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0_Q.json -g network.c --load-mdesc /opt/st/STEdgeAI/2.0/Utilities/configs/stm32n6.mdesc --load-mpool /home/test/Work/ST/stm32ai-modelzoo-services/application_code/image_classification/STM32N6/Model/my_mpools/stm32n6-app2.mpool --save-mpool-file /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool --out-dir-prefix /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/ --all-buffers-info --no-hw-sw-parallelism --cache-maintenance --enable-virtual-mem-pools --native-float --optimization 3 --Os --Omax-ca-pipe 4 --Ocache-opt --enable-epoch-controller --output-info-file c_info.json

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

Error: machine description JSON ERROR:

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

INVALID_ARGUMENT:Unexpected end of string. Expected a value.

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

^

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

Warning: Machine description File: Machine '' does not have a version field

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

saving memory pool description "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool"

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

ERROR no fwd progress threshold exceeded (you might want to check the machine description)

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

Error: Internal scheduler error

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

<<<

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

E103(CliRuntimeError): Error calling the Neural Art compiler - ['', '', 'Error: machine description JSON ERROR:', 'INVALID_ARGUMENT:Unexpected end of string. Expected a value.', '', '^', '', "Warning: Machine description File: Machine '' does not have a version field", 'saving memory pool description "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool"', 'ERROR no fwd progress threshold exceeded (you might want to check the machine description)', '', 'Error: Internal scheduler error', '']
$ cwd: None
$ args: /opt/st/STEdgeAI/2.0/Utilities/linux/stedgeai generate --target stm32n6 -m ../../../stm32ai-modelzoo/image_classification/mobilenetv2/ST_pretrainedmodel_public_dataset/flowers/mobilenet_v2_0.35_128_fft/mobilenet_v2_0.35_128_fft_int8.tflite --output /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46 --workspace /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46 --st-neural-art default@../../application_code/image_classification/STM32N6/Model/user_neuralart.json --input-data-type uint8 --inputs-ch-position chlast --output-data-type float32
ST Edge AI Core v2.0.0-20049

PASS: 0%| | 0/82 [00:00<?, ?it/s]
PASS: 9%|▊ | 7/82 [00:00<00:01, 39.76it/s]
PASS: 13%|█▎ | 11/82 [00:00<00:02, 34.83it/s]
PASS: 24%|██▍ | 20/82 [00:00<00:01, 54.28it/s]
PASS: 32%|███▏ | 26/82 [00:00<00:01, 47.81it/s]
PASS: 39%|███▉ | 32/82 [00:00<00:01, 44.83it/s]
PASS: 45%|████▌ | 37/82 [00:00<00:01, 43.41it/s]
PASS: 51%|█████ | 42/82 [00:00<00:00, 44.66it/s]
PASS: 57%|█████▋ | 47/82 [00:01<00:00, 46.05it/s]
PASS: 65%|██████▍ | 53/82 [00:01<00:00, 46.39it/s]
PASS: 71%|███████ | 58/82 [00:01<00:00, 43.25it/s]
PASS: 77%|███████▋ | 63/82 [00:01<00:00, 39.67it/s]
PASS: 83%|████████▎ | 68/82 [00:01<00:00, 41.47it/s]
PASS: 89%|████████▉ | 73/82 [00:01<00:00, 40.27it/s]

>>>> EXECUTING NEURAL ART COMPILER

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

/opt/st/STEdgeAI/2.0/Utilities/linux/atonn -i "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0.onnx" --json-quant-file "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0_Q.json" -g "network.c" --load-mdesc "/opt/st/STEdgeAI/2.0/Utilities/configs/stm32n6.mdesc" --load-mpool "/home/test/Work/ST/stm32ai-modelzoo-services/application_code/image_classification/STM32N6/Model/my_mpools/stm32n6-app2.mpool" --save-mpool-file "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool" --out-dir-prefix "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/" --all-buffers-info --no-hw-sw-parallelism --cache-maintenance --enable-virtual-mem-pools --native-float --optimization 3 --Os --Omax-ca-pipe 4 --Ocache-opt --enable-epoch-controller --output-info-file "c_info.json"

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

>>> Shell execution has FAILED (returned code = 1)

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

$ /opt/st/STEdgeAI/2.0/Utilities/linux/atonn -i /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0.onnx --json-quant-file /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/mobilenet_v2_0.35_128_fft_int8_OE_3_1_0_Q.json -g network.c --load-mdesc /opt/st/STEdgeAI/2.0/Utilities/configs/stm32n6.mdesc --load-mpool /home/test/Work/ST/stm32ai-modelzoo-services/application_code/image_classification/STM32N6/Model/my_mpools/stm32n6-app2.mpool --save-mpool-file /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool --out-dir-prefix /home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/ --all-buffers-info --no-hw-sw-parallelism --cache-maintenance --enable-virtual-mem-pools --native-float --optimization 3 --Os --Omax-ca-pipe 4 --Ocache-opt --enable-epoch-controller --output-info-file c_info.json

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

Error: machine description JSON ERROR:

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

INVALID_ARGUMENT:Unexpected end of string. Expected a value.

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

^

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

Warning: Machine description File: Machine '' does not have a version field

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

saving memory pool description "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool"

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

ERROR no fwd progress threshold exceeded (you might want to check the machine description)

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

Error: Internal scheduler error

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

<<<

PASS: 89%|████████▉ | 73/82 [00:02<00:00, 40.27it/s]

E103(CliRuntimeError): Error calling the Neural Art compiler - ['', '', 'Error: machine description JSON ERROR:', 'INVALID_ARGUMENT:Unexpected end of string. Expected a value.', '', '^', '', "Warning: Machine description File: Machine '' does not have a version field", 'saving memory pool description "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/experiments_outputs/2025_04_01_11_43_46/neural_art__network/stm32n6-app2.mpool"', 'ERROR no fwd progress threshold exceeded (you might want to check the machine description)', '', 'Error: Internal scheduler error', '']
Error executing job with overrides: []
Traceback (most recent call last):
File "/home/test/anaconda3/envs/st_zoo/lib/python3.10/site-packages/clearml/binding/hydra_bind.py", line 230, in _patched_task_function
return task_function(a_config, *a_args, **a_kwargs)
File "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/stm32ai_main.py", line 407, in main
process_mode(mode=mode, configs=cfg, train_ds=train_ds, valid_ds=valid_ds, quantization_ds=quantization_ds,
File "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/stm32ai_main.py", line 302, in process_mode
deploy(cfg=configs)
File "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/../deployment/deploy.py", line 90, in deploy
stm32ai_deploy_stm32n6(target=board, stlink_serial_number=stlink_serial_number, stm32ai_version=stm32ai_version, c_project_path=c_project_path,
File "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/../../common/deployment/common_deploy.py", line 469, in stm32ai_deploy_stm32n6
stmaic_local_call(session)
File "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/../../common/deployment/common_deploy.py", line 443, in stmaic_local_call
stmaic.compile(session=session, options=opt, target=session._board_config)
File "/home/test/Work/ST/stm32ai-modelzoo-services/image_classification/src/../../common/stm32ai_local/compile.py", line 208, in cmd_compile
raise Exception('Error during compilation')
Exception: Error during compilation

Can anyone help?

[Julian E.]

Hello @diama13,

 

Can you send me your file: deployment_n6_mobilenet_v2_config.yaml to make sure everything is setup correctly.

 

Have a good day,

Julian

[diama13]

Hi Julian,

I sent you the file by email. Also i copy/paste here the file "deployment_n6_mobilenet_v2_config.yaml":

general:
  # path to a `.tflite` or `.onnx` file.
  model_path: ../../../stm32ai-modelzoo/image_classification/mobilenetv2/ST_pretrainedmodel_public_dataset/flowers/mobilenet_v2_0.35_128_fft/mobilenet_v2_0.35_128_fft_int8.tflite

operation_mode: deployment

dataset:
  name: flowers_dataset
  class_names: [daisy, dandelion, roses, sunflowers, tulips]

preprocessing:
  resizing:
    interpolation: bilinear
    aspect_ratio: crop
  color_mode: rgb # rgb, bgr

tools:
  stedgeai:
    version: 10.0.0
    optimization: balanced
    on_cloud: False # Not Available For STM32N6
# path_to_stedgeai: C:/Users/<XXXXX>/STM32Cube/Repository/Packs/STMicroelectronics/X-CUBE-AI/<*.*.*>/Utilities/windows/stedgeai.exe
# path_to_cubeIDE: C:/ST/STM32CubeIDE_<*.*.*>/STM32CubeIDE/stm32cubeide.exe
    path_to_stedgeai: /opt/st/STEdgeAI/2.0/Utilities/linux/stedgeai
  path_to_cubeIDE: /opt/st/stm32cubeide_1.18.0/stm32cubeide

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

hydra:
  run:
    dir: ./experiments_outputs/${now:%Y_%m_%d_%H_%M_%S}

mlflow:
  uri: ./experiments_outputs/mlruns

[diama13]

Hi Julian,

I ran as superuser the command :

./stedgeai generate --model mobilenet_v2_0.35_128_fft_int8.tflite --target stm32n6 --st-neural-art

as you said and i attach the network_generate_report.txt which seems to be normal. Also, i attach the output of the terminal (terminal_output.txt) which says that "Shell execution has FAILED (returned code = 1) .... etc."

Can you explain me please why i have problems  using the deployment_n6_mobilenet_v2_config.yaml which i also attch (i added the .txt extension ir order to upload it)?

Also, what about the error in terminal_output.txt ?

Thanks in advance.

[Julian E.]

Hello @diama13 ,

 

I replicated your situation on a new install on linux.

The first time I ran the stm32ai_main.py I get the error: Exception: Error during compilation

 

To fix it, I use chmod:

sudo chmod 777 -R /opt/ST/STEdgeAI/ 
sudo chmod 777 -R /opt/ST/STEdgeAI/*

 

When I launch it again, I don't get the error anymore, but the script blocks and never end.

 

To solve that, I ran manually a generate with the command:

./stedgeai generate --model MY_MODEL --target stm32n6 --st-neural-art

And at the end of it, when running the ST Edge AI Core the first time, it asks:

Do you allow statistics to improve the command line ? (y)es / (n)

And wait util it gets an answer => which is why the model zoo do not finish.

 

After saying yes or no, the message never appears again.

So now, when I use the model zoo stm32ai_main.py, it works.

 

In your case, you did use the ST Edge AI Core manually, So I think that you don't have an issue anymore.

Please let me know.

 

Have a good day,

Julian

[diama13]

Hi Julian,

 

indeed the problem was the rights of the STEdgeAI. I managed to run the .yaml file as well. 

Thanks a lot

Have a good day.