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.

PROCESSOR-SDK-AM62A: Getting the output class in an interval

Part Number: PROCESSOR-SDK-AM62A

Tool/software:

Hello,

How to modify the output frames and implement inference that averages recognized classes across multiple frames or relates them to a time interval.

Best Regards,
Sajan

  • Hi Sajan,

    I have routed your query to our expert. Please expect a response early next week.

    Also, could you elaborate more on the use-case you are trying to execute. 

    Best Regards,

    Suren

  • Hello Suren,

    could you elaborate more on the use-case you are trying to execute. 

    When running the app_edgeai.py file. The system need only process limited frames (Reduction of frame count). And also I don't need the inference at every instant. The class with highest confidence level when checking many frames should appeared in recognized class


    Best Regards,
    Sajan

  • Hi Sajan,

    You will need to modify the source code in apps_python/infer_pipe.py.

    See the main function 'pipeline'. This function will pull the input tensor, process a frame with the model, pull the source image, postprocess, and provide an output image back to the gstreamer pipeline. If you look at the gstreamer pipeline, this is all happening between the appsrc and appsink plugins. 

    You can modify this as needed, e.g. add a new member to the local InferPipe class to track across multiple frames.

    BR,
    Reese