Error in development Object Detection using the given example

[Hood7319]

Hello,

I want to deploy a pre-trained object detection model on an STM32 board using STM32Cube.AI.

I deploy the model [ssd_mobilenet_v2_fpnlite_035_192_int8.tflite] pretrained on COCO dataset using the necessary parameters provided in [ssd_mobilenet_v2_fpnlite_035_192_config.yaml] as the example, but I got the following errors:

I only change the deployment_config.yaml.

deployment_config.yaml.txt

[INFO] : Successfully connected! [INFO] : Starting the model memory footprints estimation... Error executing job with overrides: [] Traceback (most recent call last): File "E:\Desktop\stm32ai-modelzoo-main\object_detection\src\stm32ai_main.py", line 234, in main process_mode(cfg, File "E:\Desktop\stm32ai-modelzoo-main\object_detection\src\stm32ai_main.py", line 113, in process_mode deploy(cfg) File "E:\Desktop\stm32ai-modelzoo-main\object_detection\src../deployment\deploy.py", line 98, in deploy stm32ai_deploy(target=board, stlink_serial_number=stlink_serial_number, stm32ai_version=stm32ai_version, c_project_path=c_project_path, File "E:\Desktop\stm32ai-modelzoo-main\object_detection\src../../common/deployment\common_deploy.py", line 378, in stm32ai_deploy dispatch_weights(internalFlashSizeFlash_KB=board.config.internalFlash_size, File "E:\Desktop\stm32ai-modelzoo-main\object_detection\src../../common/deployment\common_deploy.py", line 58, in dispatch_weights sorted_weights = dict(sorted(graph["weights"].items(), key=lambda item: item[1]['buffer_c_count'], reverse=True)) File "E:\Desktop\stm32ai-modelzoo-main\object_detection\src../../common/deployment\common_deploy.py", line 58, in sorted_weights = dict(sorted(graph["weights"].items(), key=lambda item: item[1]['buffer_c_count'], reverse=True)) KeyError: 'buffer_c_count'

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.

How can I sovle the problem?

Thank you for your guidance.

[Vietsol-AnhHaiLe]

I have the same problem

[h-arthi]

Hey I'm having the same issue, anyone have any luck on figuring it out?

[LFOSTM]

Hello, Could you make a try using the latest version of the model zoo, and pointing on stedgeai version 10.0.0, please? Thanks & Regards,

[woshuhua]

Hello, Could you make a try using the latest version of the model zoo, and pointing on stedgeai version 10.0.0, please? Thanks & Regards,

I am using the latest version of the model zoo and pointing on st edge AI version 10.0.0. I am getting the same error too... here's the error: [INFO] : Successfully connected! [INFO] : Starting the model memory footprints estimation... Error executing job with overrides: [] Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function [INFO] : Starting the model memory footprints estimation... Error executing job with overrides: [] Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function Error executing job with overrides: [] Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function return task_function(a_config, *a_args, **a_kwargs) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 228, in main return task_function(a_config, *a_args, **a_kwargs) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 228, in main File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 228, in main process_mode(cfg) process_mode(cfg) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 102, in process_mode deploy(cfg) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../deployment\deploy.py", line 101, in deploy stm32ai_deploy(target=board, stlink_serial_number=stlink_serial_number, stm32ai_version=stm32ai_version, c_project_path=c_project_path, File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../../common/deployment\common_deploy.py", line 378, in stm32ai_deploy dispatch_weights(internalFlashSizeFlash_KB=board.config.internalFlash_size, File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../../common/deployment\common_deploy.py", line 58, in dispatch_weights sorted_weights = dict(sorted(graph["weights"].items(), key=lambda item: item[1]['buffer_c_count'], reverse=True)) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../../common/deployment\common_deploy.py", line 58, in sorted_weights = dict(sorted(graph["weights"].items(), key=lambda item: item[1]['buffer_c_count'], reverse=True)) KeyError: 'buffer_c_count'

deployment_config.yaml.txt

[woshuhua]

Hello, Could you make a try using the latest version of the model zoo, and pointing on stedgeai version 10.0.0, please? Thanks & Regards,

I am using the latest version of the model zoo and pointing on st edge AI version 10.0.0. I am getting the same error too... here's the error: [INFO] : Successfully connected! [INFO] : Starting the model memory footprints estimation... Error executing job with overrides: [] Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function [INFO] : Starting the model memory footprints estimation... Error executing job with overrides: [] Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function Error executing job with overrides: [] Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function Traceback (most recent call last): File "C:\Users\xuhua\AppData\Local\Programs\Python\Python310\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function return task_function(a_config, *a_args, **a_kwargs) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 228, in main return task_function(a_config, *a_args, **a_kwargs) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 228, in main File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 228, in main process_mode(cfg) process_mode(cfg) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src\stm32ai_main.py", line 102, in process_mode deploy(cfg) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../deployment\deploy.py", line 101, in deploy stm32ai_deploy(target=board, stlink_serial_number=stlink_serial_number, stm32ai_version=stm32ai_version, c_project_path=c_project_path, File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../../common/deployment\common_deploy.py", line 378, in stm32ai_deploy dispatch_weights(internalFlashSizeFlash_KB=board.config.internalFlash_size, File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../../common/deployment\common_deploy.py", line 58, in dispatch_weights sorted_weights = dict(sorted(graph["weights"].items(), key=lambda item: item[1]['buffer_c_count'], reverse=True)) File "C:\Users\xuhua\Documents\stm32ai-modelzoo-services\object_detection\src../../common/deployment\common_deploy.py", line 58, in sorted_weights = dict(sorted(graph["weights"].items(), key=lambda item: item[1]['buffer_c_count'], reverse=True)) KeyError: 'buffer_c_count'

deployment_config.yaml.txt

This issue is disappeared when I change the model to st_ssd_mobilenet_v1_025_192_int8.tflite. I am not sure about the model extraction have any problem that caused the error buffer_c_count.

[GHORTESSTM]

Hello,

The issue has been solved in the last version: https://github.com/STMicroelectronics/stm32ai-modelzoo-services. You fetch the last version.

The problem was present when using the dev cloud. Now it works both with stedgeai in local and with the dev cloud.