def __init__(
self,
summary_writer=None,
interval=1,
log_dir="./runs",
tag_name="val_dice",
compute_metric=True,
images=True,
image_interval=1,
max_channels=1,
max_frames=64,
add_scalar=True,
merge_scalar=False,
fold_size=0,
):
self.writer = SummaryWriter(
log_dir=log_dir) if summary_writer is None else summary_writer
self.interval = interval
self.tag_name = tag_name
self.compute_metric = compute_metric
self.images = images
self.image_interval = image_interval
self.max_channels = max_channels
self.max_frames = max_frames
self.add_scalar = add_scalar
self.merge_scalar = merge_scalar
self.fold_size = fold_size
self.logger = logging.getLogger(__name__)
if torch.distributed.is_initialized():
self.tag_name = "{}-r{}".format(self.tag_name,
torch.distributed.get_rank())
self.plot_data = {}
self.metric_data = {}
"""
"""## Download dataset
Downloads and extracts the dataset.
The dataset comes from http://medicaldecathlon.com/.
"""
md5 = "410d4a301da4e5b2f6f86ec3ddba524e"
root_dir = "//home//imoreira//Data"
#root_dir = "C:\\Users\\isasi\\Downloads"
data_dir = os.path.join(root_dir, "Kidneys_Data")
out_dir = os.path.join(root_dir, "Kidneys_Best_Model")
tensorboard_dir = "//home//imoreira//Data//Tensorboard_Kidneys"
writer = SummaryWriter(log_dir=tensorboard_dir)
"""## Set MSD Spleen dataset path"""
train_images = sorted(glob.glob(os.path.join(data_dir, "imagesTr",
"*.nii.gz")))
train_labels = sorted(glob.glob(os.path.join(data_dir, "labelsTr",
"*.nii.gz")))
data_dicts = [{
"image": image_name,
"label": label_name
} for image_name, label_name in zip(train_images, train_labels)]
#n = len(data_dicts)
#train_files, val_files = data_dicts[:-3], data_dicts[-3:]
#train_files, val_files = data_dicts[:int(n*0.8)], data_dicts[int(n*0.2):]
val_files, train_files, test_files = data_dicts[0:8], data_dicts[