def get_arguments():
parser = argparse.ArgumentParser(description='Tree segmentation')
# Model parameters
parser.add_argument(
'--nfeatures',
type=int,
default=8,
help='number of features present in the first layer of the network')
parser.add_argument('--input',
default='/storage/workspace/dtd/images/scaly',
help='input image path')
parser.add_argument('--display',
action='store_true',
default=False,
help='display progress of generated images')
parser.add_argument(
'--image_size',
type=int,
default=32,
help='the height / width of the input image to network')
parser.add_argument('--beta1',
type=float,
default=0.5,
help='beta1 for adam. default=0.5')
parser.add_argument('--output', default="output", help="output path")
common.add(parser)
models.add_arguments(parser)
return parser.parse_args()
def _get_arguments(argv):
parser = argparse.ArgumentParser()
config.add_arguments(parser)
models.add_arguments(parser)
solver.add_arguments(parser)
loss_metrics.add_arguments(parser)
input_pipeline.add_arguments(parser)
custom_evaluator.add_arguments(parser)
args = parser.parse_args(argv[1:])
config.check_args(args, parser)
config.fill_default_args(args)
return args
'--train_path',
default='train.hdf5',
help='Input path for pre-processed training data and labels')
parser.add_argument(
'--val_path',
default='val.hdf5',
help='Input path for pre-processed validation data and labels')
parser.add_argument('--output_path',
default='model.hdf5',
help='Output path for model weights')
parser.add_argument('--log_path',
default='logs/UCSD/upscale/',
help='Output path for TensorFlow logs')
# model specification
parser = models.add_arguments(parser)
# training
parser.add_argument(
'--gpu_frac',
type=float,
default=0.,
help='Fraction of GPU memory to allocate (TensorFlow only)')
parser.add_argument(
'--tile_size',
type=int,
default='-1',
help='Tile size: -1 for no tiling, 1 for patches, n>1 for nxn tiles')
parser.add_argument('--batch_size', type=int, default=32, help='Batch size')
parser.add_argument('--batches_per_epoch',
type=int,