batch_size=2,
shuffle=True,
pin_memory=True)
valid_loader = torch.utils.data.DataLoader(valid_dataset,
batch_size=2,
shuffle=True,
pin_memory=True)
data_root = r'D:\liver2\liver2'
lr = 3e-5
x_train_dir = os.path.join(data_root, 'train-150')
y_train_dir = os.path.join(data_root, 'train/masks')
x_test_dir = os.path.join(data_root, 'test/imgs')
y_test_dir = os.path.join(data_root, 'test/masks')
loss_unet = loss.DiceLoss(weight=0.2,
activation='softmax2d',
ignore_channels=[0]) + loss.FocalLoss()
optimizer_unet = torch.optim.Adam(unet.parameters(), lr=lr)
metrics = [
metrics.SMPIoU(threshold=0.5, ignore_channels=[0], activation='softmax2d'),
metrics.Fscore(threshold=0.5, ignore_channels=[0], activation='softmax2d'),
]
train_epoch = run.TrainEpoch(
model=unet,
loss=loss_unet,
metrics=metrics,
optimizer=optimizer_unet,
device=DEVICE,
verbose=True,
)
r"D:\liver2\liver2\test\imgs",
r"D:\liver2\liver2\test\masks",
augmentation=pet_augmentation_valid(),
preprocessing=get_preprocessing(preproc_fn),
classes=['tissue', 'pancreas'],
maxsize=99999
)
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=2, shuffle=True)
valid_loader = torch.utils.data.DataLoader(valid_dataset, batch_size=2, shuffle=True)
data_root = r'D:\liver2\liver2'
lr = 3e-5
x_train_dir = os.path.join(data_root, 'train-150')
y_train_dir = os.path.join(data_root, 'train/masks')
x_test_dir = os.path.join(data_root, 'test/imgs')
y_test_dir = os.path.join(data_root, 'test/masks')
loss_unet = loss.DiceLoss(weight=0.2, activation='softmax2d', ignore_channels=[0]) + loss.FocalLoss()
optimizer_unet = torch.optim.Adam(unet.parameters(), lr=lr)
metrics = [
metrics.SMPIoU(threshold=0.5, ignore_channels=[0], activation='softmax2d'),
metrics.Fscore(threshold=0.5, ignore_channels=[0], activation='softmax2d'),
]
train_epoch = run.TrainEpoch(
model=unet,
loss=loss_unet,
metrics=metrics,
optimizer=optimizer_unet,
device=DEVICE,
verbose=True,
)
