def define_G(input_nc, output_nc, ngf, n_downsample_global=3, n_blocks_global=9, norm='instance'):
norm_layer = get_norm_layer(norm_type=norm)
netG = GlobalGenerator(input_nc, output_nc, ngf, n_downsample_global, n_blocks_global, norm_layer)
#assert (torch.cuda.is_available())
netG.cuda() #cuda Azade
netG.apply(weights_init)
return netG
def define_D(input_nc, ndf, n_layers_D, norm='instance', use_sigmoid=False, num_D=1):
norm_layer = get_norm_layer(norm_type=norm)
netD = MultiscaleDiscriminator(input_nc, ndf, n_layers_D, norm_layer, use_sigmoid, num_D)
# print(netD)
assert (torch.cuda.is_available()) # #cuda Azade
netD.cuda() #cuda Azade
netD.apply(weights_init)
return netD
def define_mask_D(input_nc,
ndf,
n_layers_D,
norm='instance',
use_sigmoid=False,
num_D=1,
num_objects=None):
norm_layer = get_norm_layer(norm_type=norm)
netD = MultiscaleDiscriminator_2(input_nc, ndf, n_layers_D, norm_layer,
use_sigmoid, num_D, num_objects)
assert (torch.cuda.is_available())
netD.cuda()
netD.apply(weights_init)
return netD