def data_preprare(self, test=False):
list_data = get_files(self.data_dir, test)
if test:
test_dataset = dataset(list_data=list_data,
test=True,
transform=data_transforms(
'val', self.normalise_type))
return test_dataset
else:
data_pd = pd.DataFrame({
"data": list_data[0],
"label": list_data[1]
})
train_pd, val_pd = train_test_split_order(data_pd,
test_size=0.2,
num_classes=12)
train_dataset = dataset(list_data=train_pd,
transform=data_transforms(
'train', self.normalise_type))
val_dataset = dataset(list_data=val_pd,
transform=data_transforms(
'val', self.normalise_type))
print(len(train_dataset))
print(len(val_dataset))
return train_dataset, val_dataset
def data_preprare(self, test=False):
list_data = get_files(self.data_dir, test)
if test:
test_dataset = dataset(list_data=list_data,
test=True,
transform=data_transforms(
'val', self.normalise_type))
return test_dataset
else:
data_pd = pd.DataFrame({
"data": list_data[0],
"label": list_data[1]
})
train_pd, val_pd = train_test_split(data_pd,
test_size=0.20,
random_state=40,
stratify=data_pd["label"])
train_dataset = dataset(list_data=train_pd,
transform=data_transforms(
'train', self.normalise_type))
val_dataset = dataset(list_data=val_pd,
transform=data_transforms(
'val', self.normalise_type))
return train_dataset, val_dataset
def data_split(self, transfer_learning=True):
if transfer_learning:
# get source train and val
list_data = get_files(self.data_dir, self.source_N)
data_pd = pd.DataFrame({"data": list_data[0], "label": list_data[1]})
train_pd, val_pd = train_test_split(data_pd, test_size=0.2, random_state=40, stratify=data_pd["label"])
source_train = dataset(list_data=train_pd, transform=self.data_transforms['train'])
source_val = dataset(list_data=val_pd, transform=self.data_transforms['val'])
# get target train and val
list_data = get_files(self.data_dir, self.target_N)
data_pd = pd.DataFrame({"data": list_data[0], "label": list_data[1]})
train_pd, val_pd = train_test_split(data_pd, test_size=0.2, random_state=40, stratify=data_pd["label"])
target_train = dataset(list_data=train_pd, transform=self.data_transforms['train'])
target_val = dataset(list_data=val_pd, transform=self.data_transforms['val'])
return source_train, source_val, target_train, target_val
else:
#get source train and val
list_data = get_files(self.data_dir, self.source_N)
data_pd = pd.DataFrame({"data": list_data[0], "label": list_data[1]})
train_pd, val_pd = train_test_split(data_pd, test_size=0.2, random_state=40, stratify=data_pd["label"])
source_train = dataset(list_data=train_pd, transform=self.data_transforms['train'])
source_val = dataset(list_data=val_pd, transform=self.data_transforms['val'])
# get target train and val
list_data = get_files(self.data_dir, self.target_N)
data_pd = pd.DataFrame({"data": list_data[0], "label": list_data[1]})
target_val = dataset(list_data=data_pd, transform=self.data_transforms['val'])
return source_train, source_val, target_val
