Search tips
Showing 42,973 results View by: Thread Post Sort by

  • PROCESSOR-SDK-TDAX: How to set the data type of model output

    Part Number: PROCESSOR-SDK-TDAX Other Parts Discussed in Thread: TDA4VH I used the officially released demo of the classification network and found that the output data type was float32. However, using quantization toolkit , the output of the model…

  • PROCESSOR-SDK-TDAX: The int8 quantization scheme has serious loss of accuracy

    TI Thinks Resolved
    Part Number: PROCESSOR-SDK-TDAX The resnet50 network uses int8 quantization, and the model accuracy drops very seriously. Then use int16 quantization scheme, and the accuracy loss is acceptable. Is there a problem with int8 quantization? The model used…

  • PROCESSOR-SDK-TDAX: The CPU usage of the classification model is too high

    Part Number: PROCESSOR-SDK-TDAX Other Parts Discussed in Thread: TDA4VH platform：TDA4VH sdk-version: 9.0 model： resnet18 link: drive.google.com/.../view Using the DL demo in the official vision_apps provided by TI to run the classified network…
    • Answered

  • PROCESSOR-SDK-TDAX: Processor SDK RTOS J721E import custom model

    Resolved
    Part Number: PROCESSOR-SDK-TDAX Hi, I have question about setup for import custom model. * I'm checking git repo to understand how I should create environment. https://github.com/TexasInstruments/edgeai-tidl-tools#setup Then, It seems that I need to install…
    • Resolved
    • Answered

  • PROCESSOR-SDK-TDAX: Processor SDK RTOS script error

    Resolved
    Part Number: PROCESSOR-SDK-TDAX Hello, Now I am trying to import my custom model(classification) by using SDK. However, when I tried to perform following script, I got following error. -Script- * run_python_examples.sh -Error- * ValueError: Failed to…
    • Resolved

  • PROCESSOR-SDK-TDAX: capture embedded metadata with the data type 0x12

    Part Number: PROCESSOR-SDK-TDAX Other Parts Discussed in Thread: TDA4VM Hi, We used the patch provided here: https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1107616/faq-processor-sdk-j721e-how-to-capture-embedded-metadata-with…

  • PROCESSOR-SDK-TDAX: An error occurred when import yolov5s onnx model

    Part Number: PROCESSOR-SDK-TDAX env: EVM: TDA4VH sdk: 9.0 LESSOPEN=| /usr/bin/lesspipe %s TF_REPO_PATH=/opt/ti-processor-sdk-rtos-j784s4-evm-09_00_00_02/tensorflow TIDL_PROTOBUF_PATH=/opt/ti-processor-sdk-rtos-j784s4-evm-09_00_00_02/protobuf…
    • Answered

  • PROCESSOR-SDK-TDAX: PyTIOVX - Kernel.setParameter() data_types not supported for Type.VX_TENSOR

    Resolved
    Part Number: PROCESSOR-SDK-TDAX Hi, I just realized that support for Type.VX_TENSOR is missing when using the data_types argument of Kernel.setParameter(), unlike all other types (see kernel_code.py -> generate_host_c_validate_func_code() Is…
    • Resolved
    • Answered

  • [FAQ] PROCESSOR-SDK-TDAX: How to change input display pipeline in the DSS M2M WB node?

    Resolved
    Part Number: PROCESSOR-SDK-TDAX The patches shared on below link adds support for DSS memory to memory path in the vision apps and TIOVX. These patches are also integrated SDK8.6. By default, this patch uses VIDL2 pipeline as input pipeline and WB pipeline…
    • Resolved
    • Answered

  • PROCESSOR-SDK-TDAX: tivxMapTensorPatch() fails when passing NULL to optional parameters

    Resolved
    Part Number: PROCESSOR-SDK-TDAX Title says it all. I call: tivxMapTensorPatch(ctx.mat1 , nDims, nullptr, nullptr, &mapId , stride, (void**) &data , VX_WRITE_ONLY, VX_MEMORY_TYPE_HOST); And the function returns VX_ERROR_INVALID_PARAMETERS…
    • Resolved

Didn't find what you are looking for? Post a new question.