def init():
base_path = os.path.dirname(__file__)
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
dd = pdb.set_trace
try:
os.makedirs(os.path.join(base_path, args.outf))
except OSError:
pass
print(torch.cuda.is_available())
if torch.cuda.is_available() and not args.cuda:
print(
"WARNING: You have a CUDA device, so you should probably run with --cuda"
)
G_xvz = _G_xvz()
G_vzx = _G_vzx()
train_list = os.path.join(base_path, args.data_list)
train_loader = torch.utils.data.DataLoader(data_loader_evaluate.ImageList(
train_list,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])),
batch_size=args.batch_size,
shuffle=False,
num_workers=args.workers,
pin_memory=True)
x = torch.FloatTensor(args.batch_size, 3, 128, 128)
x_bar_bar_out = torch.FloatTensor(10, 3, 128, 128)
v_siz = 9
z_siz = 128 - v_siz
v = torch.FloatTensor(args.batch_size, v_siz)
z = torch.FloatTensor(args.batch_size, z_siz)
if args.cuda:
G_xvz = torch.nn.DataParallel(G_xvz).cuda()
G_vzx = torch.nn.DataParallel(G_vzx).cuda()
x = x.cuda()
x_bar_bar_out = x_bar_bar_out.cuda()
v = v.cuda()
z = z.cuda()
x = Variable(x)
x_bar_bar_out = Variable(x_bar_bar_out)
v = Variable(v)
z = Variable(z)
load_pretrained_model(G_xvz, os.path.join(base_path, args.modelf),
'netG_xvz.pth')
load_pretrained_model(G_vzx, os.path.join(base_path, args.modelf),
'netG_vzx.pth')
batch_size = args.batch_size
cudnn.benchmark = True
G_xvz.eval()
G_vzx.eval()
return x, v, z, x_bar_bar_out, G_xvz, G_vzx, batch_size, train_loader
args = parser.parse_args()
print(args)
try:
os.makedirs(args.outf)
except OSError:
pass
if torch.cuda.is_available() and not args.cuda:
print(
"WARNING: You have a CUDA device, so you should probably run with --cuda"
)
# need initialize!!
G_xvz = _G_xvz()
G_vzx = _G_vzx()
D_xvs = _D_xvs()
G_xvz.apply(weights_init)
G_vzx.apply(weights_init)
D_xvs.apply(weights_init)
train_list = args.data_list
train_loader = torch.utils.data.DataLoader(data_loader.ImageList(
train_list,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])),
batch_size=args.batch_size,
args = parser.parse_args()
print(args)
try:
os.makedirs(args.outf)
except OSError:
pass
if torch.cuda.is_available() and not args.cuda:
print(
"WARNING: You have a CUDA device, so you should probably run with --cuda"
)
# need initialize!!
G_xvz_random = _G_xvz()
G_xvz_SP = _G_xvz()
G_vzx_random = _G_vzx()
G_vzx_SP = _G_vzx()
train_list = args.data_list
train_loader = torch.utils.data.DataLoader(data_loader_evaluate.ImageList(
train_list,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])),
batch_size=args.batch_size,
shuffle=False,
num_workers=args.workers,