def __restore(self, file_name, sess):
variables = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope="network/mobilenet_encoder")
dict = load_obj(file_name)
for variable in variables:
for key, value in dict.items():
print('Layer Loaded ', key)
if key in variable.name:
sess.run(tf.assign(variable, value))
def __restore(file_name, sess):
variables = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES,
scope="network/mobilenet_encoder")
net_dict = load_obj(file_name)
for variable in variables:
# 去掉network/
name_in_pkl = variable.name[8:]
if name_in_pkl in net_dict:
print('Loaded:', variable.name)
sess.run(tf.assign(variable, net_dict[name_in_pkl]))
def __convert_graph_names(self, path):
"""
This function is to convert from the mobilenet original model pretrained weights structure to our
model pretrained weights structure.
:param path: (string) path to the original pretrained weights .pkl file
:return: None
"""
dict = load_obj(path)
variables = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='mobilenet_encoder')
dict_output = {}
# for variable in variables:
# print(variable.name)
# for key, value in dict.items():
# print(key)
for key, value in dict.items():
for variable in variables:
for i in range(len(dict)):
for j in range(len(variables)):
if ((key.find("Conv2d_" + str(i) + "_") != -1 and variable.name.find(
"conv_ds_" + str(j) + "/") != -1) and i + 1 == j):
if key.find("depthwise") != -1 and variable.name.find(
"depthwise") != -1 and (key.find("gamma") != -1 and variable.name.find(
"gamma") != -1 or key.find("beta") != -1 and variable.name.find(
"beta") != -1) or key.find("pointwise") != -1 and variable.name.find(
"pointwise") != -1 and (key.find("gamma") != -1 and variable.name.find(
"gamma") != -1 or key.find("beta") != -1 and variable.name.find(
"beta") != -1) or key.find("pointwise/weights") != -1 and variable.name.find(
"pointwise/weights") != -1 or key.find(
"depthwise_weights") != -1 and variable.name.find(
"depthwise/weights") != -1 or key.find("pointwise/biases") != -1 and variable.name.find(
"pointwise/biases") != -1 or key.find("depthwise/biases") != -1 and variable.name.find(
"depthwise/biases") != -1 or key.find("1x1/weights") != -1 and variable.name.find(
"1x1/weights") != -1 or key.find("1x1/biases") != -1 and variable.name.find(
"1x1/biases") != -1:
dict_output[variable.name] = value
elif key.find(
"Conv2d_0/") != -1 and variable.name.find("conv_1/") != -1:
if key.find("weights") != -1 and variable.name.find("weights") != -1 or key.find(
"biases") != -1 and variable.name.find(
"biases") != -1 or key.find("beta") != -1 and variable.name.find(
"beta") != -1 or key.find("gamma") != -1 and variable.name.find(
"gamma") != -1:
dict_output[variable.name] = value
save_obj(dict_output, self.pretrained_path)
print("Pretrained weights converted to the new structure. The filename is mobilenet_v1.pkl.")
def __restore(self, file_name, sess):
variables = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES)
try:
print("Loading ImageNet pretrained weights...")
dict = load_obj(file_name)
run_list = []
for variable in variables:
for key, value in dict.items():
# Adding ':' means that we are interested in the variable itself and not the variable parameters
# that are used in adaptive optimizers
if key + ":" in variable.name:
run_list.append(tf.assign(variable, value))
sess.run(run_list)
print("ImageNet Pretrained Weights Loaded Initially\n\n")
except KeyboardInterrupt:
print("No pretrained ImageNet weights exist. Skipping...\n\n")
def __save(self, sess):
file_name= self.pretrained_path
variables = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES)
try:
print("Loading ImageNet pretrained weights...")
dict = load_obj(file_name)
saved_dict= {}
run_list = []
for variable in variables:
for key, value in dict.items():
if key + ":" in variable.name:
saved_dict[key]= sess.run(variable)
save_obj(saved_dict, 'pretrained_weights/shufflenet_cityscapes.pkl')
for k, v in saved_dict.items():
if k not in dict.keys():
import pdb; pdb.set_trace()
print("ImageNet Pretrained Weights Loaded Initially\n\n")
except KeyboardInterrupt:
print("No pretrained ImageNet weights exist. Skipping...\n\n")
def prepare_dataloaders(dataset_split,
dataset_path,
metadata_filename,
batch_size=32,
sample_size=-1,
valid_split=0.8):
'''
Utility function to prepare dataloaders for training.
Parameters
----------
dataset_split : str
Any of 'train', 'extra', 'test'.
dataset_path : str
Absolute path to the dataset. (i.e. .../data/SVHN/train')
metadata_filename : str
Absolute path to the metadata pickle file.
batch_size : int
Mini-batch size.
sample_size : int
Number of elements to use as sample size,
for debugging purposes only. If -1, use all samples.
valid_split : float
Returns a validation split of %size; valid_split*100,
valid_split should be in range [0,1].
Returns
-------
if dataset_split in ['train', 'extra']:
train_loader: torch.utils.DataLoader
Dataloader containing training data.
valid_loader: torch.utils.DataLoader
Dataloader containing validation data.
if dataset_split in ['test']:
test_loader: torch.utils.DataLoader
Dataloader containing test data.
'''
assert dataset_split in ['train', 'test', 'extra'], "check dataset_split"
metadata = load_obj(metadata_filename)
# dataset_path = datadir / dataset_split
firstcrop = FirstCrop(0.3)
rescale = Rescale((64, 64))
random_crop = RandomCrop((54, 54))
to_tensor = ToTensor()
# Declare transformations
transform = transforms.Compose(
[firstcrop, rescale, random_crop, to_tensor])
dataset = SVHNDataset(metadata, data_dir=dataset_path, transform=transform)
indices = np.arange(len(metadata))
# indices = np.random.permutation(indices)
# Only use a sample amount of data
if sample_size != -1:
indices = indices[:sample_size]
if dataset_split in ['train', 'extra']:
train_idx = indices[:round(valid_split * len(indices))]
valid_idx = indices[round(valid_split * len(indices)):]
train_sampler = torch.utils.data.SubsetRandomSampler(train_idx)
valid_sampler = torch.utils.data.SubsetRandomSampler(valid_idx)
# Prepare a train and validation dataloader
train_loader = DataLoader(dataset,
batch_size=batch_size,
shuffle=False,
num_workers=4,
sampler=train_sampler)
valid_loader = DataLoader(dataset,
batch_size=batch_size,
shuffle=False,
num_workers=4,
sampler=valid_sampler)
return train_loader, valid_loader
elif dataset_split in ['test']:
test_sampler = torch.utils.data.SequentialSampler(indices)
# Prepare a test dataloader
test_loader = DataLoader(dataset,
batch_size=batch_size,
num_workers=4,
shuffle=False,
sampler=test_sampler)
return test_loader
def prepare_dataloaders(
dataset_split,
dataset_path,
metadata_filename,
batch_size=32,
sample_size=-1,
valid_split=0.1,
test_split=0.1,
num_worker=0,
valid_metadata_filename=None,
valid_dataset_dir=None,
):
"""
Utility function to prepare dataloaders for training.
Parameters
----------
dataset_split : str
Any of 'train', 'extra', 'test'.
dataset_path : str
Absolute path to the dataset. (i.e. .../data/SVHN/train')
metadata_filename : str
Absolute path to the metadata pickle file.
batch_size : int
Mini-batch size.
sample_size : int
Number of elements to use as sample size,
for debugging purposes only. If -1, use all samples.
valid_split : float
Returns a validation split of %size; valid_split*100,
valid_split should be in range [0,1].
Returns
-------
if dataset_split in ['train', 'extra']:
train_loader: torch.utils.DataLoader
Dataloader containing training data.
valid_loader: torch.utils.DataLoader
Dataloader containing validation data.
if dataset_split in ['test']:
test_loader: torch.utils.DataLoader
Dataloader containing test data.
"""
assert dataset_split in ["train", "test", "extra"], "check dataset_split"
metadata = load_obj(metadata_filename)
# dataset_path = datadir / dataset_split
firstcrop = FirstCrop(0.3)
downscale = Rescale((64, 64))
random_crop = RandomCrop((54, 54))
to_tensor = ToTensor()
normalize = None
# normalize = Normalize((0.434, 0.442, 0.473), (0.2, 0.202, 0.198))
# Declare transformations
transform = transforms.Compose(
[firstcrop, downscale, random_crop, to_tensor])
test_transform = transforms.Compose(
[FirstCrop(0.1), Rescale((54, 54)), to_tensor])
dataset = SVHNDataset(
metadata,
data_dir=dataset_path,
transform=transform,
normalize_transform=normalize,
)
dataset_length = len(metadata)
indices = np.arange(dataset_length)
# Only use a sample amount of data
if sample_size != -1:
indices = indices[:sample_size]
dataset_length = sample_size
if dataset_split in ["train", "extra"]:
# Prepare a train and validation dataloader
valid_loader = None
if valid_dataset_dir is not None:
valid_metadata = load_obj(valid_metadata_filename)
valid_dataset = SVHNDataset(
valid_metadata,
data_dir=valid_dataset_dir,
transform=test_transform,
normalize_transform=normalize,
)
valid_loader = DataLoader(
valid_dataset,
batch_size=batch_size,
shuffle=False,
num_workers=num_worker,
)
train_sampler = torch.utils.data.SubsetRandomSampler(indices)
train_loader = DataLoader(
dataset,
batch_size=batch_size,
shuffle=False,
sampler=train_sampler,
num_workers=num_worker,
)
return train_loader, valid_loader
elif dataset_split in ["test"]:
test_sampler = torch.utils.data.SequentialSampler(indices)
# change the transformer pipeline
dataset.transform = test_transform
# Prepare a test dataloader
test_loader = DataLoader(
dataset,
batch_size=batch_size,
num_workers=4,
shuffle=False,
sampler=test_sampler,
)
return test_loader
def prepare_test_dataloader(dataset_path, metadata_filename, batch_size,
sample_size):
"""
Utility function to prepare dataloaders for testing.
Parameters of the configuration (cfg)
-------------------------------------
dataset_path : str
Absolute path to the test dataset. (i.e. .../data/SVHN/test')
metadata_filename : str
Absolute path to the metadata pickle file.
batch_size : int
Mini-batch size.
sample_size : int
Number of elements to use as sample size,
for debugging purposes only. If -1, use all samples.
valid_split : float
Returns a validation split of %size; valid_split*100,
valid_split should be in range [0,1].
Returns
-------
test_loader: torch.utils.DataLoader
Dataloader containing test data.
"""
firstcrop = FirstCrop(0.3)
rescale = Rescale((64, 64))
random_crop = RandomCrop((54, 54))
to_tensor = ToTensor()
# Declare transformations
transform = transforms.Compose([
firstcrop,
rescale,
random_crop,
to_tensor,
])
# Load metadata file
metadata = load_obj(metadata_filename)
# Initialize Dataset
dataset = SVHNDataset(metadata, data_dir=dataset_path, transform=transform)
indices = np.arange(len(metadata))
# Only use a sample amount of data
if sample_size != -1:
indices = indices[:sample_size]
test_sampler = torch.utils.data.SequentialSampler(indices)
# Prepare a test dataloader
test_loader = DataLoader(dataset,
batch_size=batch_size,
num_workers=4,
shuffle=False,
sampler=test_sampler)
return test_loader
def prepare_dataloaders(cfg):
dataset_path = cfg.INPUT_DIR
metadata_filename = cfg.METADATA_FILENAME
batch_size = cfg.TRAIN.BATCH_SIZE
sample_size = cfg.TRAIN.SAMPLE_SIZE
valid_split = cfg.TRAIN.VALID_SPLIT
"""
Utility function to prepare dataloaders for training.
Parameters of the configuration (cfg)
-------------------------------------
dataset_path : str
Absolute path to the dataset. (i.e. .../data/SVHN/train')
metadata_filename : str
Absolute path to the metadata pickle file.
batch_size : int
Mini-batch size.
sample_size : int
Number of elements to use as sample size,
for debugging purposes only. If -1, use all samples.
valid_split : float
Returns a validation split of %size; valid_split*100,
valid_split should be in range [0,1].
Returns
-------
train_loader: torch.utils.DataLoader
Dataloader containing training data.
valid_loader: torch.utils.DataLoader
Dataloader containing validation data.
"""
firstcrop = FirstCrop(0.3)
rescale = Rescale((64, 64))
random_crop = RandomCrop((54, 54))
to_tensor = ToTensor()
# Declare transformations
transform = transforms.Compose([
firstcrop,
rescale,
random_crop,
to_tensor,
])
# index 0 for train subset, 1 for extra subset
metadata_train = load_obj(metadata_filename[0])
metadata_extra = load_obj(metadata_filename[1])
train_data_dir = dataset_path[0]
extra_data_dir = dataset_path[1]
valid_split_train = valid_split[0]
valid_split_extra = valid_split[1]
# Initialize the combined Dataset
dataset = FullSVHNDataset(metadata_train,
metadata_extra,
train_data_dir,
extra_data_dir,
transform=transform)
indices_train = np.arange(len(metadata_train))
indices_extra = np.arange(len(metadata_train),
len(metadata_extra) + len(metadata_train))
# Only use a sample amount of data
if sample_size[0] != -1:
indices_train = indices_train[:sample_size[0]]
if sample_size[1] != -1:
indices_extra = indices_extra[:sample_size[1]]
# Select the indices to use for the train/valid split from the 'train' subset
train_idx_train = indices_train[:round(valid_split_train *
len(indices_train))]
valid_idx_train = indices_train[round(valid_split_train *
len(indices_train)):]
# Select the indices to use for the train/valid split from the 'extra' subset
train_idx_extra = indices_extra[:round(valid_split_extra *
len(indices_extra))]
valid_idx_extra = indices_extra[round(valid_split_extra *
len(indices_extra)):]
# Combine indices from 'train' and 'extra' as one single train/validation split
train_idx = np.concatenate((train_idx_train, train_idx_extra))
valid_idx = np.concatenate((valid_idx_train, valid_idx_extra))
# Define the data samplers
train_sampler = torch.utils.data.SubsetRandomSampler(train_idx)
valid_sampler = torch.utils.data.SubsetRandomSampler(valid_idx)
# Prepare a train and validation dataloader
train_loader = DataLoader(dataset,
batch_size=batch_size,
shuffle=False,
num_workers=4,
sampler=train_sampler)
valid_loader = DataLoader(dataset,
batch_size=batch_size,
shuffle=False,
num_workers=4,
sampler=valid_sampler)
return train_loader, valid_loader