def prepare(self, priming_data):
if self._prepared:
raise Exception(
'You can only call "prepare" once for a given encoder.')
self._model = NnEncoderHelper(images)
self._prepared = True
class NnAutoEncoder:
def __init__(self, is_target=False):
self._model = None
self._pytorch_wrapper = torch.FloatTensor
self._prepared = False
def prepare_encoder(self, priming_data):
if self._prepared:
raise Exception(
'You can only call "prepare_encoder" once for a given encoder.'
)
self._model = NnEncoderHelper(images)
self._prepared = True
def encode(self, images):
"""
Encode all the images from the list of paths(to images)
:param images: List of images paths
:return: a torch.floatTensor
"""
if not self._prepared:
raise Exception(
'You need to call "prepare_encoder" before calling "encode" or "decode".'
)
if not self._model:
logging.error("No model to encode, please train the model")
return self._model.encode(images)
def decode(self, encoded_values_tensor, save_to_path="decoded/"):
"""
Decoded the encoded list of image tensors and write the decoded images to give path
:param encoded_values_tensor: List of encoded images tensors
:param save_to_path: Path to store decoded images
:return: a list of image paths
"""
if not self._model:
logging.error("No model to decode, please train the model")
if not os.path.exists(save_to_path):
os.makedirs(save_to_path)
return self._model.decode(encoded_values_tensor, save_to_path)
def train(self, images):
"""
:param images: List of images paths
"""
self._model = NnEncoderHelper(images)
class NnAutoEncoder:
def __init__(self, images, is_target = False):
self._model = NnEncoderHelper(images)
self._pytorch_wrapper = torch.FloatTensor
def encode(self, images):
"""
Encode all the images from the list of paths(to images)
:param images: List of images paths
:return: a torch.floatTensor
"""
if not self._model:
logging.error("No model to encode, please train the model")
return self._model.encode(images)
def decode(self, encoded_values_tensor, save_to_path="decoded/"):
"""
Decoded the encoded list of image tensors and write the decoded images to give path
:param encoded_values_tensor: List of encoded images tensors
:param save_to_path: Path to store decoded images
:return: a list of image paths
"""
if not self._model:
logging.error("No model to decode, please train the model")
if not os.path.exists(save_to_path):
os.makedirs(save_to_path)
return self._model.decode(encoded_values_tensor, save_to_path)
def train(self, images):
"""
:param images: List of images paths
"""
self._model = NnEncoderHelper(images)
def train(self, images):
"""
:param images: List of images paths
"""
self._model = NnEncoderHelper(images)
def __init__(self, images, is_target = False):
self._model = NnEncoderHelper(images)
self._pytorch_wrapper = torch.FloatTensor