torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
print(args)
# Get all patient ids
patient_ids = get_patient_ids('../dataset/', 400)
print(len(patient_ids))
for domain in range(10):
train_patient_ids = patient_ids[:]
test_patient_ids = patient_ids[domain]
train_patient_ids.remove(test_patient_ids)
train_dataset = MalariaData('../dataset/',
domain_list=train_patient_ids,
transform=True)
train_size = int(0.80 * len(train_dataset))
test_size = len(train_dataset) - train_size
train_dataset, val_dataset = torch.utils.data.random_split(
train_dataset, [train_size, test_size])
train_loader = data_utils.DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
**kwargs)
val_loader = data_utils.DataLoader(val_dataset,
batch_size=args.batch_size,
shuffle=True,
**kwargs)
print(args)
# Train, val, test sets
patient_ids = get_patient_ids('../dataset/', 400)
print(len(patient_ids))
train_patient_ids = patient_ids[:]
test_patient_ids = patient_ids[0]
train_patient_ids_unsupervised = 'C59P20'
train_patient_ids.remove(test_patient_ids)
train_patient_ids.remove(train_patient_ids_unsupervised)
print(test_patient_ids, train_patient_ids, train_patient_ids_unsupervised)
train_dataset = MalariaData('../dataset/', domain_list=train_patient_ids, transform=True)
train_size = int(0.80 * len(train_dataset))
test_size = len(train_dataset) - train_size
train_dataset, val_dataset = torch.utils.data.random_split(train_dataset, [train_size, test_size])
train_loader_supervised = data_utils.DataLoader(train_dataset, batch_size=args.batch_size, shuffle=True,
**kwargs)
val_loader_supervised = data_utils.DataLoader(val_dataset, batch_size=args.batch_size, shuffle=True, **kwargs)
test_loader = data_utils.DataLoader(
MalariaData('../dataset/', domain_list=[test_patient_ids]),
batch_size=1,
shuffle=False,
**kwargs)
train_loader_unsupervised = data_utils.DataLoader(train_dataset, batch_size=args.batch_size, shuffle=True,
model.qy.activation = nn.LeakyReLU()
batch_size = 14
# seeds
torch.manual_seed(1)
np.random.seed(1)
patient_ids = get_patient_ids('../dataset/', 400)
print(len(patient_ids))
train_patient_ids = patient_ids[:]
test_patient_ids = patient_ids[0]
train_patient_ids.remove(test_patient_ids)
train_dataset = MalariaData('../dataset/', domain_list=train_patient_ids)
train_size = int(0.80 * len(train_dataset))
test_size = len(train_dataset) - train_size
train_dataset, val_dataset = torch.utils.data.random_split(
train_dataset, [train_size, test_size])
train_loader = data_utils.DataLoader(train_dataset,
batch_size=batch_size,
shuffle=True,
**kwargs)
# Get 8 images
for batch_idx, (x, y, d) in enumerate(train_loader):
with torch.no_grad():
x, y, d = x.cuda(), y.cuda(), d.cuda()
break