Hi, I am having a project using predominantly acceleration data. For the training part and other stuffs, the DPS results all contain positive values, like from zero to forward. But when I deployed the custom block (.zip) in my Raspberry Pi, the acceleration value I recorded from MPU6050 sensor consist of negative values. Should I convert them to positive so that I can guarantee the accuracy from deployed tf model? Thanks in advance.
No the values will be normalised over a range you can leave them as is.
Best
Eoin