def __init__(self):
Dataset.__init__(self)
self.data_dir = VOC2007_TF_DATADIR
self.file_pattern = 'voc2007_%s.tfrecord'
self.num_classes = 20
self.name = 'VOC2007'
self.is_multilabel = True
def __init__(self, class_name):
Dataset.__init__(self)
self.data_dir = OBJECTNET3D_TF_DATADIR
self.class_name = class_name
self.num_classes = 100
self.name = 'ObjectNet3D'
self.is_multilabel = True
def __init__(self):
Dataset.__init__(self)
self.data_dir = IMAGENET_TF_DATADIR
self.file_pattern = '%s-*'
self.num_classes = 1000
self.label_offset = 1
self.name = 'imagenet'
def __init__(self):
subjects = range(1, 5)
Dataset.__init__(self,
name='OPP',
num_classes=4,
segment_length=150,
num_channels=3,
num_loc=5,
subjects=subjects)
self.XR, self.YR, self.SR = self.read_data(mode=1)
def __init__(self):
subjects = np.setdiff1d(range(1, 22),
[1, 2, 10, 12, 13, 14, 15, 17, 20, 22])
Dataset.__init__(self,
name='IRH',
num_classes=24,
segment_length=100,
num_channels=3,
num_loc=5,
subjects=subjects)
self.XR, self.YR, self.SR = self.read_data()
def __init__(self, args):
Dataset.__init__(self, args)
#pdb.set_trace()
self.user_items_dct = self.get_user_items_dict(self.user_lists_dct,
self.list_items_dct)
self.user_item_matrix_sp = self.get_sparse_matrix_from_dict(
self.user_items_dct, self.num_user, self.num_item)
self.list_item_matrix_sp = self.get_sparse_matrix_from_dict(
self.list_items_dct, self.num_list, self.num_item)
self.item_list_matrix_sp = self.get_sparse_matrix_from_dict(
self.list_items_dct, self.num_item, self.num_list, reverse=True)
t1 = time()
self.user_user_comm_mat_sp = self.mat_mult_sp(
self.user_item_matrix_sp,
self.user_item_matrix_sp.T) #.astype(bool).astype(int)#todok()
self.item_item_comm_mat_sp = self.mat_mult_sp(
self.item_list_matrix_sp,
self.item_list_matrix_sp.T) #.astype(bool).astype(int)
self.list_list_comm_mat_sp = self.mat_mult_sp(
self.list_item_matrix_sp, self.list_item_matrix_sp.T
) ##.astype(bool).astype(int) ##real values
print("""[%.2f s] """ % (time() - t1))
#pdb.set_trace()
# ==============================
#self.list_item_train_seq = self.get_dct_mat_seq(dct=self.list_items_dct, num_row=self.num_list, num_col=self.num_item, padding_value=0)
##self.train_matrix_item_seq = self.get_dct_mat_seq_remove_test(dct=self.list_items_dct, num_row=self.num_list, num_col=self.max_item_seq_length+1, padding_value=0) ##last_index :] for all, :-1] for remove test item, :-2] for removing test and valid
#pdb.set_trace()
##
'''
binarize = True
print("hello")
if binarize == True:
self.user_user_comm_mat_sp = sp.csr_matrix(sp.csr_matrix((self.user_user_comm_mat_sp),dtype=bool),dtype=int)
self.item_item_comm_mat_sp = sp.csr_matrix(sp.csr_matrix((self.item_item_comm_mat_sp),dtype=bool),dtype=int)
self.list_list_comm_mat_sp = sp.csr_matrix(sp.csr_matrix((self.list_list_comm_mat_sp),dtype=bool),dtype=int)
#self.user_user_comm_mat_sp = self.binarize_sparse_matrix(self.user_user_comm_mat_sp)
#self.item_item_comm_mat_sp = self.binarize_sparse_matrix(self.item_item_comm_mat_sp)
#self.list_list_comm_mat_sp = self.binarize_sparse_matrix(self.list_list_comm_mat_sp)
print("hello 2")
'''
# adj ===========
self.user_adj_mat = self.user_user_comm_mat_sp
self.list_adj_mat = self.list_list_comm_mat_sp
self.item_adj_mat = self.item_item_comm_mat_sp
print("hello")
def __init__(self, args):
Dataset.__init__(self, args)
self.user_item_embed_mat = self.get_user_item_embed_mat(
self.user_attr_mat, self.attr_mat)
#self.adjacency_mat = self.get_adjacency_matrix(self.train_matrix,self.train_matrix.T)
self.adjacency_mat = self.get_adjacency_matrix_sparse(
self.train_matrix, self.train_matrix.T)
assert self.adjacency_mat.shape[0] == self.adjacency_mat.shape[1]
self.num_nodes = self.adjacency_mat.shape[0]
self.feature_dim = self.attr_dim
print("num nodes: ", self.num_nodes)
def __init__(self, id=None, drawing=None,
posX=0, posY=0,
x1=0, y1=0, x2=0, y2=0,
pen=None, brush=None):
Dataset.__init__(self, id)
self.drawing = drawing
self.posX = posX
self.posY = posY
self.x1 = x1
self.y1 = y1
self.x2 = x2
self.y2 = y2
self.pen = pen
self.brush = brush
def __init__(self, data_path, train_val_test=(0.8, 0.1, 0.1)):
"""
Initialises a ImSeg_Dataset object by calling the superclass initialiser.
The difference between an ImSeg_Dataset object and a Dataset object is the annotation.
This object will therefore override the self.annotations_path and
self.annotation_list attributes.
"""
assert (train_val_test[0] + train_val_test[1] + train_val_test[2]
) == 1, 'Train, val and test percentages should add to 1'
assert train_val_test[0] > 0 and train_val_test[
1] > 0 and train_val_test[
2] > 0, 'Train, val and test percentages should be non-negative'
Dataset.__init__(self, data_path)
self.train_val_test = train_val_test
self.train_path = self.data_path + '/im_seg/train'
self.val_path = self.data_path + '/im_seg/val'
self.test_path = self.data_path + '/im_seg/test'
self.out_path = self.data_path + '/im_seg/out'
self.data_sizes = [] # [train_size, val_size, test_size, out_size]
if not os.path.isdir(self.data_path + '/im_seg'):
print(
f"Creating directory to store semantic segmentation formatted dataset."
)
os.mkdir(self.data_path + '/im_seg')
# Create train, validation, test directories, each with an images and
# annotations sub-directories
for directory in [
self.train_path, self.val_path, self.test_path, self.out_path
]:
if not os.path.isdir(directory):
os.mkdir(directory)
if not os.path.isdir(directory + '/images'):
os.mkdir(directory + '/images')
if not os.path.isdir(directory + '/annotations'):
os.mkdir(directory + '/annotations')
# Size of each training, val and test directories
num_samples = len([
name for name in os.listdir(f'{directory}/images')
if name.endswith('.jpg')
])
self.data_sizes.append(num_samples)
def __init__(self,
data_path,
train_val_test=(0.8, 0.1, 0.1),
is_plot=False):
"""
Initialises a 'PIXOR_Dataset' object by calling the superclass initialiser.
The difference between a PIXOR_Dataset object and a Dataset object is the annotation.
The PIXOR_Dataset object will therefore override the self.annotations_path and
self.annotation_list attributes such that the building labels are in XML format.
"""
assert (train_val_test[0] + train_val_test[1] + train_val_test[2]
) == 1, 'Train, val and test percentages should add to 1'
assert train_val_test[0] > 0 and train_val_test[
1] > 0 and train_val_test[
2] > 0, 'Train, val and test percentages should be non-negative'
Dataset.__init__(self, data_path)
self.train_val_test = train_val_test
self.train_path = self.data_path + '/pixor/train'
self.val_path = self.data_path + '/pixor/val'
self.test_path = self.data_path + '/pixor/test'
self.is_plot = is_plot
if not os.path.isdir(self.data_path + '/pixor'):
print(f"Creating directory to store PIXOR formatted dataset.")
os.mkdir(self.data_path + '/pixor')
# Create train, validation, test directories, each with an images and annotations
# sub-directory
for directory in [self.train_path, self.val_path, self.test_path]:
if not os.path.isdir(directory):
os.mkdir(directory)
if not os.path.isdir(directory + '/images'):
os.mkdir(directory + '/images')
if not os.path.isdir(directory + '/class_annotations'):
os.mkdir(directory + '/class_annotations')
if not os.path.isdir(directory + '/box_annotations'):
os.mkdir(directory + '/box_annotations')
def __init__(self):
subjects = range(1, 9)
Dataset.__init__(self,
name='SAD',
num_classes=19,
segment_length=125,
num_channels=3,
num_loc=5,
subjects=subjects)
if not os.path.exists(self.data_folder):
os.makedirs(self.data_folder)
self.write_to_file()
x1 = np.genfromtxt(self.name + '_allX1.csv', delimiter=',')
x2 = np.genfromtxt(self.name + '_allX2.csv', delimiter=',')
x3 = np.genfromtxt(self.name + '_allX3.csv', delimiter=',')
self.YR = np.genfromtxt(self.name + '_allY.csv', delimiter=',')
self.SR = np.genfromtxt(self.name + '_allS.csv', delimiter=',')
a, b = x1.shape
self.XR = np.empty(shape=(a, 3, b))
self.XR[:, 0, :] = x1
self.XR[:, 1, :] = x2
self.XR[:, 2, :] = x3
def __init__(self, args):
Dataset.__init__(self, args)
self.same_entity_list = eval(args.same_entity)
assert len(
self.same_entity_list
) == args.num_views, 'length of same_entity and num_views should be same.'
self.item_view_matrix,self.num_row,self.num_col,self.adjacency_view_matrix = [],[],[],[]
for view in range(args.num_views):
num_row, num_col = self.get_row_column_count(self.embed_path +
".view_matrix" +
str(view + 1))
self.num_row.append(
num_row
) # this is only for test purpose. self.num_item is used instead
self.num_col.append(num_col)
if self.same_entity_list[
view] == -1: #-1 means entities are different
self.item_view_matrix.append(
self.load_rating_file_as_matrix_for_views(
self.embed_path + ".view_matrix" + str(view + 1),
self.num_row[view], self.num_col[view]))
self.adjacency_view_matrix.append(
self.get_adjacency_matrix_sparse(
self.item_view_matrix[view], self.
item_view_matrix[view].T)) # Note num_items is used
else: # 1 or other number means it is same entity both side
self.item_view_matrix.append(
self.load_rating_file_as_matrix_for_views(
self.embed_path + ".view_matrix" + str(view + 1),
max(self.num_row[view], self.num_col[view]),
max(self.num_row[view], self.num_col[view])))
_A_obs = self.item_view_matrix[view] + self.item_view_matrix[
view].T # Note num_items is used
_A_obs[_A_obs > 1] = 1
self.adjacency_view_matrix.append(
_A_obs) # Note num_items is used
def __init__(self):
Dataset.__init__(self)
self.data_dir = STL10_TF_DATADIR
self.file_pattern = 'stl10_%s.tfrecord'
self.num_classes = 10
self.name = 'STL10'
def __init__(self, id=None, x=0, y=0):
Dataset.__init__(self, id)
self.x = x
self.y = y
