def main(argv):
(opts, args) = parser.parse_args(argv)
torch.manual_seed(opts.seed)
torch.cuda.manual_seed(opts.seed)
if not os.path.exists(opts.output_folder):
os.makedirs(opts.output_folder)
# Load experiment setting
config = get_config(opts.config)
style_dim = config['gen']['style_dim']
opts.num_style = 1 if opts.style != '' else opts.num_style
# Setup model and data loader
trainer = MUNIT_Trainer(config)
state_dict = torch.load(opts.checkpoint)
trainer.gen_a.load_state_dict(state_dict['a'])
trainer.gen_b.load_state_dict(state_dict['b'])
trainer.cuda()
trainer.eval()
encode = trainer.gen_a.encode if opts.a2b else trainer.gen_b.encode # encode function
style_encode = trainer.gen_b.encode if opts.a2b else trainer.gen_a.encode # encode function
decode = trainer.gen_b.decode if opts.a2b else trainer.gen_a.decode # decode function
transform = transforms.Compose([
transforms.Resize(config['new_size']),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])
image = Variable(transform(Image.open(
opts.input).convert('RGB')).unsqueeze(0).cuda(),
volatile=True)
style_image = Variable(transform(Image.open(
opts.style).convert('RGB')).unsqueeze(0).cuda(),
volatile=True) if opts.style != '' else None
# Start testing
style_rand = Variable(torch.randn(opts.num_style, style_dim, 1, 1).cuda(),
volatile=True)
content, _ = encode(image)
if opts.style != '':
_, style = style_encode(style_image)
else:
style = style_rand
for j in range(opts.num_style):
s = style[j].unsqueeze(0)
outputs = decode(content, s)
outputs = (outputs + 1) / 2.
path = os.path.join(opts.output_folder, 'output{:03d}.jpg'.format(j))
vutils.save_image(outputs.data, path, padding=0, normalize=True)
if not opts.output_only:
# also save input images
vutils.save_image(image.data,
os.path.join(opts.output_folder, 'input.jpg'),
padding=0,
normalize=True)
def main(argv):
(opts, args) = parser.parse_args(argv)
torch.manual_seed(opts.seed)
torch.cuda.manual_seed(opts.seed)
if not os.path.exists(opts.output_folder):
os.makedirs(opts.output_folder)
# Load experiment setting
config = get_config(opts.config)
input_dim = config['new_size'] if opts.a2b else config['input_dim_b']
style_dim = config['gen']['style_dim']
# Setup model and data loader
data_loader = get_data_loader_folder(opts.input_folder,
1,
False,
input_dim == 1,
crop=False)
trainer = MUNIT_Trainer(config)
state_dict = torch.load(opts.checkpoint)
trainer.gen_a.load_state_dict(state_dict['a'])
trainer.gen_b.load_state_dict(state_dict['b'])
trainer.cuda()
trainer.eval()
encode = trainer.gen_a.encode if opts.a2b else trainer.gen_b.encode # encode function
decode = trainer.gen_b.decode if opts.a2b else trainer.gen_a.decode # decode function
# Start testing
style_fixed = Variable(torch.randn(opts.num_style, style_dim, 1, 1).cuda(),
volatile=True)
for i, images in enumerate(data_loader):
images = Variable(images.cuda(), volatile=True)
content, _ = encode(images)
style = style_fixed if opts.synchronized else Variable(
torch.randn(opts.num_style, style_dim, 1, 1).cuda(), volatile=True)
for j in range(opts.num_style):
s = style[j].unsqueeze(0)
outputs = decode(content, s)
outputs = (outputs + 1) / 2.
path = os.path.join(opts.output_folder,
'input{:03d}_output{:03d}.jpg'.format(i, j))
vutils.save_image(outputs.data, path, padding=0, normalize=True)
if not opts.output_only:
# also save input images
vutils.save_image(images.data,
os.path.join(opts.output_folder,
'input{:03d}.jpg'.format(i)),
padding=0,
normalize=True)
style_dim = config['gen']['style_dim']
trainer = MUNIT_Trainer(config)
else:
sys.exit("Only support MUNIT")
try:
state_dict = torch.load(opts.checkpoint)
trainer.gen_a.load_state_dict(state_dict['a'])
trainer.gen_b.load_state_dict(state_dict['b'])
except:
state_dict = pytorch03_to_pytorch04(torch.load(opts.checkpoint), opts.trainer)
trainer.gen_a.load_state_dict(state_dict['a'])
trainer.gen_b.load_state_dict(state_dict['b'])
trainer.cuda()
trainer.eval()
encode = trainer.gen_a.encode if opts.a2b else trainer.gen_b.encode # encode function
style_encode = trainer.gen_b.encode if opts.a2b else trainer.gen_a.encode # encode function
decode = trainer.gen_b.decode if opts.a2b else trainer.gen_a.decode # decode function
if 'new_size' in config:
new_size = config['new_size']
else:
if opts.a2b==1:
new_size = config['new_size_a']
else:
new_size = config['new_size_b']
transform = transforms.Compose([transforms.Resize(new_size),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
config['vgg_model_path'] = opts.output_path
if opts.trainer == 'MUNIT':
style_dim = config['gen']['style_dim']
trainer = MUNIT_Trainer(config)
elif opts.trainer == 'UNIT':
trainer = UNIT_Trainer(config)
else:
sys.exit("Only support MUNIT|UNIT")
state_dict = torch.load(opts.checkpoint)
trainer.gen_a.load_state_dict(state_dict['a'])
trainer.gen_b.load_state_dict(state_dict['b'])
trainer.cuda()
trainer.eval()
encode = trainer.gen_a.encode if opts.a2b else trainer.gen_b.encode # encode function
decode = trainer.gen_b.decode if opts.a2b else trainer.gen_a.decode # decode function
if opts.trainer == 'MUNIT':
# Start testing
style_fixed = Variable(torch.randn(opts.num_style, style_dim, 1, 1).cuda(), volatile=True)
for i, (images, names) in enumerate(zip(data_loader,image_names)):
print(names[1])
images = Variable(images.cuda(), volatile=True)
content, _ = encode(images)
style = style_fixed if opts.synchronized else Variable(torch.randn(opts.num_style, style_dim, 1, 1).cuda(), volatile=True)
for j in range(opts.num_style):
s = style[j].unsqueeze(0)
outputs = decode(content, s)
outputs = (outputs + 1) / 2.
