I need your help to resolve this matter please.
In the company where I work, we are implementing a license plate recognition project with Luxonis cameras.
we have a model trained in YOLOV5 in onnx format and we need to convert it to a blob myriadx blob format to run inside oak-1.
The problem when we try to use the luxonis converter (http://blobconverter.luxonis.com/) gives us this error that I copy the error below:

Conversion error
×Close
Error message
Command failed with exit code 1, command: /opt/intel/openvino2022_1/tools/compile_tool/compile_tool -m /tmp/blobconverter/4f2843a5694c4f9ba0fc4313673db94d/best_bad/FP16/best_bad.xml -o /tmp/blobconverter/4f2843a5694c4f9ba0fc4313673db94d/best_bad/FP16/best_bad.blob -c /tmp/blobconverter/4f2843a5694c4f9ba0fc4313673db94d/myriad_compile_config.txt -d MYRIAD -ip U8
Console output (stdout)
OpenVINO Runtime version ......... 2022.1.0
Build ........... 2022.1.0-7019-cdb9bec7210-releases/2022/1
Network inputs:
images : u8 / [...]
Network outputs:
output/sink_port_0 : f16 / [...]
Error output (stderr)
[ GENERAL_ERROR ]
/home/jenkins/agent/workspace/private-ci/ie/build-linux-ubuntu20/b/repos/openvino/src/plugins/intel_myriad/graph_transformer/src/frontend/frontend.cpp:592 Failed to compile layer "onnx::Split_414": unsupported layer type "ScatterNDUpdate"

I have two files. The model that works is trained with three epochs, and the one that doesn't work is trained with 100 epochs.
I share with you the one drive where the two file folders are located.
https://brunoschillig-my.sharepoint.com/:f:/g/personal/fernandog_schillig_com_ar/EglYTsQ_JwtOsA4e9H4ql98B_qDjzzSVOHY30TII0eGUEw?e=PWGqBZ
what I need to do to resolve this problem?
Thanks.
best,
Fernando Granada.

  • erik replied to this.

    Hi fernandogranada ,
    Please follow the yolov5_training training & deployment notebook here. After the training it goes through the steps of converting, compiling and deploying the model to the OAK camera.
    Thanks, Erik

      Hi Erik, we already have the model trained with yolov5 and saved in onnx format, but when we want to convert it to myriadx blob we get this error that I copy below.

      Conversion error
      ×Close
      Error message
      Command failed with exit code 1, command: /opt/intel/openvino2022_1/tools/compile_tool/compile_tool -m /tmp/blobconverter/4f2843a5694c4f9ba0fc4313673db94d/best_bad/FP16/best_bad.xml -o /tmp/blobconverter/4f2843a5694c4f9ba0fc4313673db94d/best_bad/FP16/best_bad.blob -c /tmp/blobconverter/4f2843a5694c4f9ba0fc4313673db94d/myriad_compile_config.txt -d MYRIAD -ip U8
      Console output (stdout)
      OpenVINO Runtime version ......... 2022.1.0
      Build ........... 2022.1.0-7019-cdb9bec7210-releases/2022/1
      Network inputs:
      images : u8 / [...]
      Network outputs:
      output/sink_port_0 : f16 / [...]
      Error output (stderr)
      [ GENERAL_ERROR ]
      /home/jenkins/agent/workspace/private-ci/ie/build-linux-ubuntu20/b/repos/openvino/src/plugins/intel_myriad/graph_transformer/src/frontend/frontend.cpp:592 Failed to compile layer "onnx::Split_414": unsupported layer type "ScatterNDUpdate"

      Why does this error appear?

      • erik replied to this.

        Hi fernandogranada ,
        I'd suggest using that notebook and skipping the training part and only focusing on the conversion & deployment part. Thoughts?
        Thanks, Erik