def get_train_val_test_sureface(self, data_set_path_train,
label_set_path_train, data_set_path_test,
label_set_path_test, split_size, device):
print("Dataset Size")
train_data_set, labels_set = self.__read_dataset(
data_set_path_train, label_set_path_train)
X_train, X_val, Y_train, Y_val = self.__spilt_data_set(
train_data_set, labels_set, split_size=split_size)
X_test, Y_test = self.__read_dataset(data_set_path_test,
label_set_path_test)
print("Train set:")
print(Y_train.shape[0])
print("Val set:")
print(Y_val.shape[0])
print("Test set:")
print(Y_test.shape[0])
train_set = Util.convert_to_tensor(X_train, Y_train, device)
val_set = Util.convert_to_tensor(X_val, Y_val, device)
test_set = Util.convert_to_tensor(X_test, Y_test, device)
return train_set, Y_train.shape[0], val_set, Y_val.shape[
0], test_set, Y_test.shape[0]
def __split_train_test_validation_set_texture(self, data_set_path, label_set_path, split_size,
device):
print("Texture Dataset Size")
train_data_set, labels_set = self.__read_dataset(data_set_path, label_set_path)
self.texture_set_size = labels_set.shape[0]
X_train, X_val, Y_train, Y_val = self.__spilt_data_set(train_data_set,
labels_set,
split_size=split_size)
train_set = Util.convert_to_tensor(X_train, Y_train, device)
val_set = Util.convert_to_tensor(X_val, Y_val, device)
return train_set, val_set, train_data_set, labels_set
def get_tensor_set(self, dataset_path, label_set_path, device):
train_data_set, labels_set = self.__read_dataset(
dataset_path, label_set_path)
texture_data_set_size = labels_set.shape[0]
train_set = Util.convert_to_tensor(train_data_set, labels_set, device)
return train_set, texture_data_set_size
def pre_process_test_texture(self, data_set_path, label_set_path, device):
test_data, labels_set = self.__read_dataset(data_set_path,
label_set_path)
processed_dataset = Util.convert_to_tensor(test_data, labels_set,
device)
self.texture_set_size = labels_set.shape[0]
return processed_dataset
def split_train_test_validation_set_image_net(self, data_set_path,
label_set_path, device):
"""
This method splits the data set into train, test and validation set. Also this method resize the images
based on image dimensions specified by image_dims parameter.
:param data_set_path:
:param label_set_path:
:param split_size:
:param device:
:param flag:
:return train, test and validation set and their corresponding sizes
"""
print("ImageNet Dataset Size")
train_data_set, labels_set = self.__read_dataset(
data_set_path, label_set_path)
train_set_size = labels_set.shape[0]
# X_train, X_val, Y_train, Y_val = self.__spilt_data_set(train_data_set,
# labels_set,
# split_size=split_size)
train_set = Util.convert_to_tensor(train_data_set, labels_set, device)
# val_set = Util.convert_to_tensor(X_val, Y_val, device)
return train_set, train_set_size
