def build_input_data(input_size):
if FLAGS.validation_files.endswith('.csv'):
if FLAGS.dataset_base_dir is None:
raise RuntimeError('To use CSV files as input, you must specify'
' --dataset_base_dir')
input_data = dataloader.CSVInputProcessor(
csv_file=FLAGS.validation_files,
data_dir=FLAGS.dataset_base_dir,
batch_size=BATCH_SIZE,
is_training=False,
output_size=input_size,
num_classes=FLAGS.num_classes,
provide_filename=True,
)
else:
input_data = dataloader.TFRecordWBBoxInputProcessor(
file_pattern=FLAGS.validation_files,
batch_size=BATCH_SIZE,
is_training=False,
output_size=input_size,
num_classes=FLAGS.num_classes,
num_instances=0,
provide_filename=True,
)
dataset, _, _ = input_data.make_source_dataset()
return dataset
def _build_dataset():
if FLAGS.input_size is None:
raise RuntimeError('To use a representative dataset, you must specify'
' --input_size')
if FLAGS.representative_dataset.endswith('.csv'):
if FLAGS.dataset_base_dir is None:
raise RuntimeError('To use CSV files as input, you must specify'
' --dataset_base_dir')
input_data = dataloader.CSVInputProcessor(
csv_file=FLAGS.representative_dataset,
data_dir=FLAGS.dataset_base_dir,
batch_size=BATCH_SIZE,
is_training=False,
output_size=FLAGS.input_size,
resize_with_pad=FLAGS.resize_with_pad,
num_classes=FLAGS.num_classes,
)
else:
input_data = dataloader.TFRecordWBBoxInputProcessor(
file_pattern=FLAGS.representative_dataset,
batch_size=BATCH_SIZE,
is_training=False,
output_size=FLAGS.input_size,
resize_with_pad=FLAGS.resize_with_pad,
num_classes=FLAGS.num_classes,
num_instances=0,
)
dataset, _, _ = input_data.make_source_dataset()
return dataset
def build_tfrecord_input_data(file_pattern, num_instances, is_training=False):
if FLAGS.num_classes is None:
raise RuntimeError('To use TFRecords as input, you must specify'
' --num_classes')
input_data = dataloader.TFRecordWBBoxInputProcessor(
file_pattern=file_pattern,
batch_size=FLAGS.batch_size,
num_classes=FLAGS.num_classes,
num_instances=num_instances,
is_training=is_training,
output_size=FLAGS.input_size,
randaug_num_layers=FLAGS.randaug_num_layers,
randaug_magnitude=FLAGS.randaug_magnitude,
seed=FLAGS.random_seed,
)
return input_data.make_source_dataset()
def build_input_data():
include_geo_data = FLAGS.geo_prior_ckpt_dir is not None
input_data = dataloader.TFRecordWBBoxInputProcessor(
file_pattern=FLAGS.test_files,
batch_size=FLAGS.batch_size,
is_training=False,
output_size=FLAGS.input_size,
num_classes=FLAGS.num_classes,
num_instances=0,
provide_validity_info_output=include_geo_data,
provide_coord_date_encoded_input=include_geo_data,
provide_instance_id=True,
provide_coordinates_input=FLAGS.use_coordinates_inputs)
dataset, _, _ = input_data.make_source_dataset()
return dataset