This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

J721EXSOMXEVM: EdgeAI example program is giving "KeyError: 0"

Part Number: J721EXSOMXEVM
Other Parts Discussed in Thread: TDA4VM

Hello team,

I followed the procedure recommended by you at Part 6: Putting it all together: end-to-end AI application development but I am getting below error. Nothing like SoC env var is given in the documentation. Could you provide some solution to it?

  • Hi Kishan,

    To set SOC environment variable on TDA4VM, you can run below mentioned command.

    export SOC=j721e

    On 8.5 linux sdk the supported config file format is slightly changed.

    flow structure should follow below mentioned format :

    flow<flow_number>:[input<input_number>,model<model_number>,output<output_number>,[pos_x,pos_y,width,height]]

    In my_configuration.yaml please consider changing to below mentioned format,

    flows:
        flow0: [input0,model0,output0,[320,180,1280,720]]

    I have created internal request to update this on edgeai academy documentation page.

  • I am still getting error as in the below screenshot:

    The update which you gave is revised in the code as given below:

    Could you please provide a workaround to these errors?

  • Hi Kishan,

    Thanks for sharing detailed error log.

    In e2e forum the spacing was not reflected correctly, apologies for that, I have edited the previous response again.

    As i see in your my_configuration.yaml file 'flow0' is not intended correctly.

    yaml files follows tab spacing for inner text inclusion.

    Please consider updated response.

    Regards,

    Pratik

  • It still not resolved completely. Throwing another error. Could you provide a workaround to the below error:

    FileNotFoundError: [Errno 2] No such file or directory: '/opt/edge_ai_apps/models/classification/TFL-CL-001-mobileNetV2/param.yaml'

    What I found is that there is no such "/opt/edge_ai_apps/models" folder. Instead by default set-up these are might be installed in model-zoo. I tried looking into that in order to change the path but landed into some binary file reference which I could not edit in editor. Below is the reference to my investigation:

  • Hi Kishan,

    On 8.5 Linux sdk, the file format is slightly changed, instead of models directory, now we have model_zoo on latest linux sdk. 

    The compiled model artifacts can be found in directory : /opt/model_zoo.

    To use mobileNetV2 for edge inferencing please consider setting path as,

    model_path: /opt/model_zoo/TVM-CL-3090-mobileNetV2-tv

    Please note, i have created internal request to update documentation.

    Regards,

    Pratik

  • Hello team,

    Now the code seems to be running fine. But I am not getting the output on the screen. Below are the screenshots for your reference.

    TeraTerm output screen

    eDP output screen

    Could you please let me know a workaround to this problem?