def extract_tpu_data(self, tf_record, random_rotation=False):
dataset = preprocessing.get_tpu_input_tensors(
1, 'nhwc', [tf_record], num_repeats=1, shuffle_records=False,
shuffle_examples=False, filter_amount=1,
random_rotation=random_rotation)
pos_tensor, label_tensors = dataset.make_one_shot_iterator().get_next()
return self.get_data_tensors(pos_tensor, label_tensors)
def _input_fn(params):
return preprocessing.get_tpu_input_tensors(
params['batch_size'],
tf_records,
filter_amount=FLAGS.filter_amount,
shuffle_examples=FLAGS.shuffle_examples,
shuffle_buffer_size=FLAGS.shuffle_buffer_size,
random_rotation=True)
def input_fn(params):
return preprocessing.get_tpu_input_tensors(params['batch_size'],
tf_records,
filter_amount=0.05)
def input_fn(params):
return preprocessing.get_tpu_input_tensors(params['batch_size'],
tf_records,
random_rotation=True)
def input_fn(params):
return preprocessing.get_tpu_input_tensors(params['batch_size'],
tf_records)
def _input_fn(params):
return preprocessing.get_tpu_input_tensors(
params['batch_size'],
tf_records, filter_amount=0.05, shuffle_examples=False)
def _input_fn(params):
return preprocessing.get_tpu_input_tensors(
params['train_batch_size'],
params['input_layout'],
tf_records,
filter_amount=1.0)
