Please help me about stargan HOT 2 CLOSED

@Gigibulid The pretrained models were trained with 5 attributes. Therefore, the dimension of generator's first conv filter is [64, 8, 7, 7], where 8 is the sum of 3(rgb channels) and 5(attributes). When you run the evaluation command with 2 attributes (black and blond), the dimension of generator's first conv filter is [64, 5, 7, 7]. This will cause a dimension mismatch error, when loading the trained parameters.

For the case of 5 attributes(black, blond, brown, male, young), You can change only hair colors(black and blond) by editing c_trg_list to c_trg_list[0:2]. Then, run the evaluation script with 5 attributes.

from stargan.

@yunjey Thank you so much!! It worked!!

How I can add my own dataset (even if that is one image) in the pre-trained model?? I changed the --dataset value from CelebA to the name of mine but it did not work.

from stargan.