• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 102 subscribers
  • Views 142 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TDA4VM: the vision app tidl output is not matching with the infer result using the PC_dsp_test_dl_algo.out tool.

meng yu
Prodigy 40 points
Part Number: TDA4VM

      I encountered the following problems when deploying the neural network model on TDA4.

     The int16 keypoint detection model,  the infer tool (PC_dsp_test_dl_algo.out) generates the bin file when doing inference. And I parses the bin file , the result is correct. The TIDL output result saved in the vision app was parsed with a numerical error and did not match the expected network output value. The TIDL output in vision app is named tidl_output_*_tensor_*_ch0.bin . The key point detection model is sensitive to numerical values, and numerical differences result in completely incorrect results in post-processing parsing.

      Why this error occurs？Please give me some guidance.

  • 0
    TI__Expert 4335 points

    Hi,

    few questions

    a) Is this your own model ?

    b) which app of vision_apps are you using?

    c) Can you please share the bin output in both the scenario?

    d) what do you mean by "a numerical error" ?

    regards

    Deepak Poddar

  • 0 in reply to Deepak Poddar
    Prodigy 40 points

    vision_app_error.zip

    a) yep,it is my own model.

    b) I am using segmentation vision app

    c) the bin output is attached, pld.bin is the infer result using the 'PC_dsp_test_dl_algo.out' tool, tidl_output*.bin is the tidl node output, and the zip file contains other useful files.

    d) a numerical error means that the vision app tidl node ouput value is not expected and I think it's wrong.