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.

SK-TDA4VM: input_mean and input_scale meaning

Genius 13655 points

Part Number: SK-TDA4VM

Hello Champs,

Customer got a wrong result when quantifying his own custom model with the default input_mean and input_scale setting, just like in the above image. He's not sure what's the meaning of the variable input_mean and input_scale. Could you help me to understanding how to calculate the parameter for custom model quantization? 


Thanks
Regards
Shine

  • Hi Shine,

         These parameters are used to normalize the input before feeding it to TIDL ( as part of pre-processing step). Following operation will be happening internally on image data using these parameters

      • image = image - input_mean 
      • image = image * input_scale

    Regards,

    Anshu