device = 'cuda'
torch.backends.cudnn.benchmark = True
transform = transforms.Compose(
[
transforms.Resize(args.size),
transforms.CenterCrop(args.size),
transforms.ToTensor(),
transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]),
]
)
loader_train, loader_val, _ = \
iPERLoader(data_root=args.path, batch=args.batch_size, transform=transform).data_load()
model = VQVAE_SPADE(embed_dim=128, parser=parser).to(device)
model = nn.DataParallel(model).cuda()
print('Loading Model_SPADE...', end='')
model.load_state_dict(torch.load('/p300/mem/mem_src/SPADE/checkpoint/as_101/vqvae_072.pt'))
model.eval()
print('Complete !')
optimizer = optim.Adam(model.parameters(), lr=args.lr)
scheduler = None
model_cond = poseVQVAE().to(device)
model_cond = nn.DataParallel(model_cond).cuda()
print('Loading Model_condition...', end='')
model_cond.load_state_dict(torch.load('/p300/mem/mem_src/checkpoint/pose_06_black/vqvae_016.pt'))
model_cond.eval()
print('Complete !')
device = 'cuda'
torch.backends.cudnn.benchmark = True
transform = transforms.Compose([
transforms.Resize(args.size),
transforms.CenterCrop(args.size),
transforms.ToTensor(),
transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]),
])
_, _, loader = iPERLoader(data_root=args.path,
batch=args.batch_size,
transform=transform).data_load()
model = VQVAE_SPADE(embed_dim=128, parser=parser).to(device)
model = nn.DataParallel(model).cuda()
# print('Loading Model...', end='')
# model.load_state_dict(torch.load('/p300/mem/mem_src/SPADE/checkpoint/app_v04/vqvae_089.pt'))
# model.eval()
# print('Complete !')
optimizer = optim.Adam(model.parameters(), lr=args.lr)
scheduler = None
model_cond = poseVQVAE().to(device)
model_cond = nn.DataParallel(model_cond).cuda()
print('Loading Model...', end='')
model_cond.load_state_dict(
torch.load('/p300/mem/mem_src/checkpoint/pose_04/vqvae_462.pt'))
model_cond.eval()
device = 'cuda'
torch.backends.cudnn.benchmark = True
transform = transforms.Compose([
transforms.Resize(args.size),
transforms.CenterCrop(args.size),
transforms.ToTensor(),
transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]),
])
_, loader, _ = iPERLoader(data_root=args.path,
batch=args.batch_size,
transform=transform).data_load()
model = VQVAE_SPADE(embed_dim=128, parser=parser).to(device)
model = nn.DataParallel(model).cuda()
print('Loading Model...', end='')
model.load_state_dict(
torch.load('/p300/mem/mem_src/SPADE/checkpoint/as_82/vqvae_244.pt'))
model.eval()
print('Complete !')
optimizer = optim.Adam(model.parameters(), lr=args.lr)
scheduler = None
model_cond = poseVQVAE().to(device)
model_cond = nn.DataParallel(model_cond).cuda()
print('Loading Model...', end='')
model_cond.load_state_dict(
torch.load('/p300/mem/mem_src/checkpoint/pose_04/vqvae_462.pt'))
device = 'cuda'
torch.backends.cudnn.benchmark = True
transform = transforms.Compose(
[
transforms.Resize(args.size),
transforms.CenterCrop(args.size),
transforms.ToTensor(),
transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]),
]
)
_, loader, _ = iPERLoader(data_root=args.path, batch=args.batch_size, transform=transform).data_load()
model = VQVAE_SPADE(embed_dim=128, parser=parser).to(device)
model = nn.DataParallel(model).cuda()
# print('Loading Model...', end='')
# model.load_state_dict(torch.load('/p300/mem/mem_src/SPADE/checkpoint/app_v04/vqvae_089.pt'))
# model.eval()
# print('Complete !')
optimizer = optim.Adam(model.parameters(), lr=args.lr)
scheduler = None
model_cond = poseVQVAE().to(device)
model_cond = nn.DataParallel(model_cond).cuda()
print('Loading Model...', end='')
model_cond.load_state_dict(torch.load('/p300/mem/mem_src/checkpoint/pose_04/vqvae_462.pt'))
model_cond.eval()
print('Complete !')
os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
device = 'cuda'
torch.backends.cudnn.benchmark = True
transform = transforms.Compose([
transforms.Resize(args.size),
transforms.CenterCrop(args.size),
transforms.ToTensor(),
transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]),
])
loader_train, loader_val, _ = \
iPERLoader(data_root=args.path, batch=args.batch_size, transform=transform).data_load()
model = VQVAE_SPADE(embed_dim=128, parser=parser).to(device)
model = nn.DataParallel(model).cuda()
print('Loading Model...', end='')
model.load_state_dict(
torch.load('/p300/mem/mem_src/SPADE/checkpoint/as_115/vqvae_014.pt'))
model.eval()
print('Complete !')
# optimizer = optim.Adam(model.parameters(), lr=args.lr)
optimizer = build_optimizer(model, lr=args.lr)
scheduler = None
model_cond = poseVQVAE().to(device)
model_cond = nn.DataParallel(model_cond).cuda()
print('Loading Model_condition...', end='')
model_cond.load_state_dict(
torch.load('/p300/mem/mem_src/checkpoint/pose_06_black/vqvae_120.pt'))