QAMFace
QAMFace copied to clipboard
Not normalised -1 to 1?
Ideally the input data is normalised between -1 and 1 because fp32 and fp16 have better precision in that range. Is there a reason why it is commented out in the code?
Hi! trans.ToTensor()
will automatically do this.
Did you test this? AFAIK, toTensor normalised to [0,1] range