def load_dataset():
global mean, std
if args.data == './data/target_128':
traindir = os.path.join(args.data, 'train')
valdir = os.path.join(args.data, 'val')
mean = [0.5, 0.5, 0.5]
std = [0.5, 0.5, 0.5]
normalize = transforms.Normalize(mean, std)
post_transforms = transforms.Compose([
transforms.ToTensor(),
normalize
])
val_transforms = transforms.Compose([
transforms.ToTensor(),
normalize,
])
train_dataset = EasyDR(traindir, None, post_transforms, args.co_power)
val_dataset = EasyDR(valdir, None, val_transforms, args.co_power)
print('load targeted easy-classified diabetic retina dataset with size 128 to pretrain unet successfully!!')
else:
raise ValueError('')
train_loader = DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=4,
drop_last=False,
pin_memory=True if args.cuda else False)
val_loader = DataLoader(val_dataset,
batch_size=args.batch_size,
shuffle=False,
drop_last=False,
num_workers=4,
pin_memory=True if args.cuda else False)
return train_loader, val_loader
def load_dataset(data_dir):
if data_dir == '../data/target_128':
mean = [0.651, 0.4391, 0.2991]
std = [0.1046, 0.0846, 0.0611]
print('load DR with 128 successfully!!!')
else:
raise ValueError(
"parameter 'data' that means path to dataset must be in "
"['./data/target_128', ./data/split_contrast_dataset]")
traindir = os.path.join(data_dir, 'train')
valdir = os.path.join(data_dir, 'val')
normalize = transforms.Normalize(mean, std)
post_transforms = transforms.Compose([transforms.ToTensor(), normalize])
train_dataset = EasyDR(traindir, None, post_transforms, alpha=0)
val_dataset = EasyDR(valdir, None, post_transforms, alpha=0)
train_loader = DataLoader(train_dataset,
batch_size=64,
shuffle=False,
num_workers=2,
pin_memory=False)
val_loader = DataLoader(val_dataset,
batch_size=64,
shuffle=False,
num_workers=2,
pin_memory=False)
return train_loader, val_loader
def init():
data = '../data/target_128'
traindir = os.path.join(data, 'train')
mean = [0.5, 0.5, 0.5]
std = [0.5, 0.5, 0.5]
normalize = transforms.Normalize(mean, std)
pre_transforms = transforms.Compose([
transforms.RandomHorizontalFlip(),
transforms.RandomVerticalFlip(),
transforms.ColorJitter(0.05, 0.05, 0.05, 0.05)
])
post_transforms = transforms.Compose([
transforms.ToTensor(),
normalize
])
train_dataset = EasyDR(traindir, pre_transforms, post_transforms, 2)
train_loader = DataLoader(train_dataset,
batch_size=1,
shuffle=True,
num_workers=1,
drop_last=True,
pin_memory=False)
unet = generator()
unet.load_state_dict(weight_to_cpu('../gan112/epoch_400/g.pkl'))
return unet, train_loader
def load_dataset():
if args.data == './data/flip':
traindir = os.path.join(args.data, 'train')
valdir = os.path.join(args.data, 'val')
mean = [0.5, 0.5, 0.5]
std = [0.5, 0.5, 0.5]
normalize = transforms.Normalize(mean, std)
# pre_transforms = transforms.Compose([
# transforms.RandomHorizontalFlip(),
# transforms.RandomVerticalFlip(),
# transforms.RandomRotation(10),
# transforms.ColorJitter(0.05, 0.05, 0.05, 0.05)
# ])
pre_transforms = transforms.Compose([
transforms.RandomRotation(5),
transforms.ColorJitter(0.05, 0.05, 0.05, 0.05)
])
post_transforms = transforms.Compose(
[transforms.ToTensor(), normalize])
val_transforms = transforms.Compose([
transforms.ToTensor(),
normalize,
])
train_dataset = EasyDR(traindir,
pre_transforms,
post_transforms,
alpha=0)
val_dataset = EasyDR(valdir, None, val_transforms, alpha=0)
print('load flipped DR successfully!!!')
else:
raise ValueError(
"parameter 'data' that means path to dataset must be in "
"['./data/target_128', './data/flip']")
train_loader = DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=4,
pin_memory=True if args.cuda else False)
val_loader = DataLoader(val_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=1,
pin_memory=True if args.cuda else False)
return train_loader, val_loader
def load_dataset():
if args.data == './data/target_128':
mean = [0.651, 0.4391, 0.2991]
std = [0.1046, 0.0846, 0.0611]
print('load DR with 128 successfully!!!')
elif args.data == './data/split_contrast_dataset':
mean = [0.7432, 0.661, 0.6283]
std = [0.0344, 0.0364, 0.0413]
print('load custom-defined skin dataset successfully!!!')
else:
raise ValueError(
"parameter 'data' that means path to dataset must be in "
"['./data/target_128', ./data/split_contrast_dataset]")
traindir = os.path.join(args.data, 'train')
valdir = os.path.join(args.data, 'val')
normalize = transforms.Normalize(mean, std)
pre_transforms = transforms.Compose([
transforms.RandomHorizontalFlip(),
transforms.RandomVerticalFlip(),
transforms.RandomRotation(10),
transforms.ColorJitter(0.05, 0.05, 0.05, 0.05)
])
post_transforms = transforms.Compose([transforms.ToTensor(), normalize])
val_transforms = transforms.Compose([
transforms.ToTensor(),
normalize,
])
train_dataset = EasyDR(traindir, pre_transforms, post_transforms, alpha=0)
val_dataset = EasyDR(valdir, None, val_transforms, alpha=0)
train_loader = DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=2,
pin_memory=True if args.cuda else False)
val_loader = DataLoader(val_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=2,
pin_memory=True if args.cuda else False)
return train_loader, val_loader
def load_dataset():
if args.data == './gan174_output':
mean = [0.7672, 0.5209, 0.3615]
std = [0.0865, 0.079, 0.0618]
print('load UNet outputs successfully!!!')
else:
raise ValueError("parameter 'data' that means path to dataset must be in "
"['./gan174_output']")
traindir = os.path.join(args.data, 'train')
valdir = os.path.join(args.data, 'val')
normalize = transforms.Normalize(mean, std)
pre_transforms = transforms.Compose([
transforms.RandomHorizontalFlip(),
transforms.RandomVerticalFlip(),
transforms.RandomRotation(10),
transforms.ColorJitter(0.05, 0.05, 0.05, 0.05)
])
post_transforms = transforms.Compose([
transforms.ToTensor(),
normalize
])
val_transforms = transforms.Compose([
transforms.ToTensor(),
normalize,
])
train_dataset = EasyDR(traindir, None, post_transforms, alpha=0)
val_dataset = EasyDR(valdir, None, val_transforms, alpha=0)
train_loader = DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=2,
pin_memory=True if args.cuda else False)
val_loader = DataLoader(val_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=2,
pin_memory=True if args.cuda else False)
return train_loader, val_loader
def get_mean_and_std(path,
transform=transforms.Compose([transforms.ToTensor()]),
channels=3):
from utils.read_data import EasyDR
dataset = EasyDR(path, None, transform)
dataloader = DataLoader(dataset,
batch_size=1,
shuffle=False,
num_workers=2)
mean = torch.zeros(channels)
std = torch.zeros(channels)
print('==> Computing mean and std..')
for inputs, targets, _, _ in dataloader:
for i in range(channels):
mean[i] += inputs[:, i, :, :].mean()
std[i] += inputs[:, i, :, :].std()
mean.div_(len(dataset))
std.div_(len(dataset))
mean, std = mean.numpy().tolist(), std.numpy().tolist()
return [round(x, 4) for x in mean], [round(y, 4) for y in std]