raspberrypi4B NCS2 Failed to deploy trained model

[Brieve]

Hello!

My environment: openvino2021.4, Buster10.13 on raspberrypi4b(32 bytes), python3.7.

 I trained a model, and converted it to IR model. However, when I ran it on raspberrypi with NCS2, I got an error. 

fight_nsc3.py:35: DeprecationWarning: BILINEAR is deprecated and will be removed in Pillow 10 (2023-07-01). Use Resampling.BILINEAR instead.
def __init__(self, size, interpolation=Image.BILINEAR):
Loading IR to the plugin...
Traceback (most recent call last):
File "fight_nsc3.py", line 209, in <module>
main()
File "fight_nsc3.py", line 141, in main
exec_net = ie.load_network(network=net, num_requests=1, device_name='MYRIAD')
File "ie_api.pyx", line 372, in openvino.inference_engine.ie_api.IECore.load_network
File "ie_api.pyx", line 390, in openvino.inference_engine.ie_api.IECore.load_network
RuntimeError: [ GENERAL_ERROR ]
/home/jenkins/agent/workspace/private-ci/ie/build-linux-debian_9_arm/b/repos/openvino/inference-engine/src/vpu/graph_transformer/src/frontend/frontend.cpp:441 Failed to compile layer "ScatterND_43": unsupported layer type "ScatterNDUpdate"

What should I do?

Thanks a lot!

[Hairul_Intel · Accepted solution]

Hi Brieve,

Thank you for reaching out to us.

For your information, the error is caused by unsupported layer (ScatterNDUpdate) on NCS2 as it is not listed in the Supported Layers.

On another note, if you're using PyTorch, do check out this workaround from GitHub thread and see if it's applicable to your model.

Regards,

Hairul

[Brieve]

Thank you very much!

My problem is exactly solved by the method you mentioned.

[Hairul_Intel]

Hi Brieve,

Glad to know that your issue has been resolved.

This thread will no longer be monitored since this issue has been resolved. If you need any additional information from Intel, please submit a new question.

Regards,

Hairul