Open bobofzhang opened 5 years ago
To discriminate if the two input images belong to the same identity, i.e., for discriminating real/fake pairs.
To discriminate if the two input images belong to the same identity, i.e., for discriminating real/fake pairs.
hi, since A and B are not belong to the same identity, for the second discriminator, there is no such "real" pairs, for training.
There are actually real A pairs involved in training but are not shown in the README figure.
hi, the first discriminator is for real/fake, what's the purpose of the second discriminator?