def define_G(opt, name='network_G'):
opt_net = opt[name]
which_model = opt_net['which_model_G']
if which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(in_nc=opt_net['in_nc'], out_nc=opt_net['out_nc'],
nf=opt_net['nf'], nb=opt_net['nb'])
elif which_model == 'RRDBNetSEG':
netG = RRDBNet_arch.RRDBNetSEG(in_nc=opt_net['in_nc'], out_nc=opt_net['out_nc'],
nf=opt_net['nf'], nb=opt_net['nb'], segm_mask=opt['train']['segm_mask'],
scale=opt_net['scale'])
else:
raise NotImplementedError('Generator model [{:s}] not recognized'.format(which_model))
return netG
def define_G(opt):
opt_net = opt['network_G']
which_model = opt_net['which_model_G']
if which_model == 'MSRResNet':
netG = SRResNet_arch.MSRResNet(in_nc=opt_net['in_nc'],
out_nc=opt_net['out_nc'],
nf=opt_net['nf'],
nb=opt_net['nb'],
upscale=opt_net['scale'],
differential=opt_net['diff'])
elif which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(in_nc=opt_net['in_nc'],
out_nc=opt_net['out_nc'],
nf=opt_net['nf'],
nb=opt_net['nb'],
differential=opt_net['diff'],
time_dependent=opt_net['time_dependent'],
adjoint=opt_net['adjoint'],
sb=opt_net['sb'])
elif which_model == 'ReacDiff':
netG = rd_net_arch.ReacDiff(in_nc=opt_net['in_nc'],
out_nc=opt_net['out_nc'],
nf=opt_net['nf'],
nb=opt_net['nb'],
differential=opt_net['diff'])
# elif which_model == 'sft_arch': # SFT-GAN
# netG = sft_arch.SFT_Net()
else:
raise NotImplementedError(
'Generator model [{:s}] not recognized'.format(which_model))
return netG
def define_G(opt):
opt_net = opt['network_G']
which_model = opt_net['which_model_G']
if which_model == 'MSRResNet':
netG = SRResNet_arch.MSRResNet(in_nc=opt_net['in_nc'], out_nc=opt_net['out_nc'],
nf=opt_net['nf'], nb=opt_net['nb'], upscale=opt_net['scale'])
elif which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(in_nc=opt_net['in_nc'], out_nc=opt_net['out_nc'],
nf=opt_net['nf'], nb=opt_net['nb'])
elif which_model == 'Predictor':
netG = sftmd_arch.Predictor()
elif which_model == 'Corrector':
netG = sftmd_arch.Corrector()
elif which_model == 'SFTMD':
netG = sftmd_arch.SFTMD()
elif which_model == 'SRResNet':
netG = sftmd_arch.SRResNet()
elif which_model == 'SFTMD_DEMO':
netG = sftmd_arch.SFTMD_DEMO()
# elif which_model == 'sft_arch': # SFT-GAN
# netG = sft_arch.SFT_Net()
else:
raise NotImplementedError('Generator model [{:s}] not recognized'.format(which_model))
return netG
def define_G(opt):
opt_net = opt["network_G"]
which_model = opt_net["which_model_G"]
if which_model == "MSRResNet":
netG = SRResNet_arch.MSRResNet(
in_nc=opt_net["in_nc"],
out_nc=opt_net["out_nc"],
nf=opt_net["nf"],
nb=opt_net["nb"],
upscale=opt_net["scale"],
)
elif which_model == "RRDBNet":
netG = RRDBNet_arch.RRDBNet(
in_nc=opt_net["in_nc"],
out_nc=opt_net["out_nc"],
nf=opt_net["nf"],
nb=opt_net["nb"],
)
elif which_model == "Predictor":
netG = sftmd_arch.Predictor(
in_nc=opt_net["in_nc"], nf=opt_net["nf"], code_len=opt_net["code_length"]
)
elif which_model == "Corrector":
netG = sftmd_arch.Corrector(
in_nc=opt_net["in_nc"], nf=opt_net["nf"], code_len=opt_net["code_length"]
)
elif which_model == "SFTMD":
netG = sftmd_arch.SFTMD(
in_nc=opt_net["in_nc"],
out_nc=opt_net["out_nc"],
nf=opt_net["nf"],
nb=opt_net["nb"],
scale=opt_net["upscale"],
input_para=opt_net["code_length"],
)
elif which_model == "SRResNet":
netG = sftmd_arch.SRResNet()
elif which_model == "SFTMD_DEMO":
netG = sftmd_arch.SFTMD_DEMO(
in_nc=opt_net["in_nc"],
out_nc=opt_net["out_nc"],
nf=opt_net["nf"],
nb=opt_net["nb"],
scale=opt_net["upscale"],
input_para=opt_net["code_length"],
)
# elif which_model == 'sft_arch': # SFT-GAN
# netG = sft_arch.SFT_Net()
else:
raise NotImplementedError(
"Generator model [{:s}] not recognized".format(which_model)
)
return netG
def define_G(opt):
opt_net = opt
which_model = opt_net.which_model_G
if which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(in_nc=opt_net.G_in_nc,
out_nc=opt_net.out_nc,
nf=opt_net.G_nf,
nb=opt_net.nb)
else:
raise NotImplementedError(
'Generator model [{:s}] not recognized'.format(which_model))
return netG
def define_G(opt):
opt_net = opt['network_G']
which_model = opt_net['which_model_G']
if which_model == 'MSRResNet':
netG = SRResNet_arch.MSRResNet(in_nc=opt_net['in_nc'], out_nc=opt_net['out_nc'],
nf=opt_net['nf'], nb=opt_net['nb'], upscale=opt_net['scale'])
elif which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(in_nc=opt_net['in_nc'], out_nc=opt_net['out_nc'],
nf=opt_net['nf'], nb=opt_net['nb'])
elif which_model == 'DWUNet':
netG = DWUNet_arch.DWUNet()
elif which_model == 'RFDNet':
netG = RFDNet_arch.RFDN()
raise NotImplementedError(
'Generator model [{:s}] not recognized'.format(which_model))
return netG
def define_G(opt):
opt_net = opt['network_G']
which_model = opt_net['which_model_G']
if which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(in_nc=opt_net['in_nc'], out_nc=opt_net['out_nc'],
nf=opt_net['nf'], nb=opt_net['nb'], scale=opt['scale'], opt=opt)
elif which_model == 'EDSRNet':
Arch = find_model_using_name(which_model)
netG = Arch(scale=opt['scale'])
elif which_model == 'rankSRGAN':
Arch = find_model_using_name(which_model)
netG = Arch(upscale=opt['scale'])
# elif which_model == 'sft_arch': # SFT-GAN
# netG = sft_arch.SFT_Net()
else:
raise NotImplementedError('Generator model [{:s}] not recognized'.format(which_model))
return netG
def define_G(opt):
opt_net = copy(opt['network_G'])
which_model = opt_net.pop('which_model_G')
if 'scale' in opt_net:
scale = opt_net.pop('scale')
if which_model == 'MSRResNet':
netG = SRResNet_arch.MSRResNet(**opt_net, upscale=scale)
elif which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(**opt_net)
elif which_model == 'USRGAN':
netG = USRGAN_arch.USRGAN(**opt_net)
elif which_model == 'USRGANLarge':
netG = USRGANLarge_arch.USRGANLarge(**opt_net)
elif which_model == 'USRGAN_conns':
netG = USRGAN_Connections_arch.USRGANLarge(**opt_net)
elif which_model == 'BOWGAN':
netG = BOWGAN_arch.BOWGAN(**opt_net)
else:
raise NotImplementedError('Generator model [{:s}] not recognized'.format(which_model))
return netG
def define_G(opt):
opt_net = opt['network_G']
which_model = opt_net['which_model_G']
if which_model == 'MSRResNet':
netG = SRResNet_arch.MSRResNet(in_nc=opt_net['in_nc'],
out_nc=opt_net['out_nc'],
nf=opt_net['nf'],
nb=opt_net['nb'],
upscale=opt_net['scale'])
elif which_model == 'RRDBNet':
netG = RRDBNet_arch.RRDBNet(in_nc=opt_net['in_nc'],
out_nc=opt_net['out_nc'],
nf=opt_net['nf'],
nb=opt_net['nb'])
elif which_model == 'MResNet':
netG = SRResNet_arch.MResNet(in_nc=opt_net['in_nc'],
out_nc=opt_net['out_nc'],
nf=opt_net['nf'],
nb=opt_net['nb'])
elif which_model == 'ResNet_alpha_beta':
netG = SRResNet_arch.ResNet_alpha_beta()
elif which_model == 'ResNet_alpha_beta_sconv':
netG = SRResNet_arch.ResNet_alpha_beta_sconv()
elif which_model == 'ResNet_alpha_beta_fc':
netG = SRResNet_arch.ResNet_alpha_beta_fc()
elif which_model == 'ResNet_alpha_beta_fc_statistics':
netG = SRResNet_arch.ResNet_alpha_beta_fc_statistics()
elif which_model == 'ResNet_alpha_beta_decoder_1x1':
netG = SRResNet_arch.ResNet_alpha_beta_decoder_1x1()
elif which_model == 'ResNet_alpha_beta_decoder_3x3':
netG = SRResNet_arch.ResNet_alpha_beta_decoder_3x3()
elif which_model == 'ResNet_alpha_beta_decoder_3x3_BN':
netG = SRResNet_arch.ResNet_alpha_beta_decoder_3x3_BN()
elif which_model == 'ResNet_alpha_beta_decoder_3x3_IN':
netG = SRResNet_arch.ResNet_alpha_beta_decoder_3x3_IN()
elif which_model == 'ResNet_alpha_beta_decoder_3x3_IN_encoder':
netG = SRResNet_arch.ResNet_alpha_beta_decoder_3x3_IN_encoder()
elif which_model == 'ResNet_alpha_beta_decoder_3x3_IN_encoder_8HW':
netG = SRResNet_arch.ResNet_alpha_beta_decoder_3x3_IN_encoder_8HW()
elif which_model == 'ResNet_alpha_beta_decoder_3x3_IN_encoder_global2local':
netG = SRResNet_arch.ResNet_alpha_beta_decoder_3x3_IN_encoder_global2local(
)
elif which_model == 'ResNet_plain':
netG = SRResNet_arch.ResNet_plain()
elif which_model == 'DFN':
netG = DynamicF.DFN_16L_2d()
elif which_model == 'DFN_1x1':
netG = DynamicF.DFN_16L_2d_1x1()
elif which_model == 'DFN_noRx':
netG = DynamicF.DFN_16L_2d(res=False)
elif which_model == 'DFN_alpha':
netG = DynamicF.DFN_16L_2d_alpha()
elif which_model == 'DCCN':
netG = DynamicF.DCCN_16L_2d()
elif which_model == 'DCCN_alpha':
netG = DynamicF.DCCN_16L_2d_alpha()
# elif which_model == 'sft_arch': # SFT-GAN
# netG = sft_arch.SFT_Net()
else:
raise NotImplementedError(
'Generator model [{:s}] not recognized'.format(which_model))
return netG