about distance loss about ceecnet HOT 2 CLOSED

thanks, i got it

from ceecnet.

Hi @cl886699 it is multiplied by 100, in order to create (and store) 100 subdivisions in the range [0,1], in uint8 format - for compression reasons. Within the dataset getitem this is fixed where first the uint8 data are made float32, and then divided by 100 so as to bring them in the correct range.

So you don't need to do anything in the loss function, the provided labels are in the range [0,1] in float32 format, from the dataset class. But please keep in mind, the source code is a demo, to understand what we did in the paper, and adopt to your needs :).

All the best,
Foivos

from ceecnet.