def cellfinder_parser():
parser = ArgumentParser(formatter_class=ArgumentDefaultsHelpFormatter)
parser = main_parse(parser)
parser = registration_parse(parser)
parser = atlas_parse(parser)
parser = pixel_parser(parser)
parser = geometry_parser(parser)
parser = cellfinder_opt_parse(parser)
parser = io_parse(parser)
parser = cell_detect_parse(parser)
parser = classification_parse(parser)
parser = cube_extract_parse(parser)
parser = count_summary_parse(parser)
parser = figures_parse(parser)
parser = amap_vis_parser(parser)
parser = standard_space_parse(parser)
parser = misc_parse(parser)
parser = amap_parser(parser)
parser = model_parser(parser)
parser = download_directory_parser(parser)
return parser
def cellfinder_parser():
parser = ArgumentParser(formatter_class=ArgumentDefaultsHelpFormatter)
parser = main_parse(parser)
parser = config_parse(parser)
parser = pixel_parser(parser)
parser = run_parse(parser)
parser = io_parse(parser)
parser = cell_detect_parse(parser)
parser = classification_parse(parser)
parser = cube_extract_parse(parser)
parser = misc_parse(parser)
parser = model_parser(parser)
parser = figures_parse(parser)
parser = download_directory_parser(parser)
# brainreg options
parser = atlas_parse(parser)
parser = geometry_parser(parser)
parser = brainreg_backend_parse(parser)
# This needs to be abstracted away into brainreg for multiple backends
parser = niftyreg_parse(parser)
return parser
def training_parse():
from cellfinder.tools.parser import misc_parse
from cellfinder.download.cli import model_parser, download_directory_parser
training_parser = ArgumentParser(
formatter_class=ArgumentDefaultsHelpFormatter)
training_parser.add_argument(
"-y",
"--yaml",
dest="yaml_file",
nargs="+",
required=True,
type=str,
help="The path to the yaml run file.",
)
training_parser.add_argument(
"-o",
"--output-dir",
dest="output_dir",
required=True,
type=str,
help="Output directory for the final model.",
)
training_parser.add_argument(
"--continue-training",
dest="continue_training",
action="store_true",
help="Continue training from an existing trained model. If no model "
"or model weights are specified, this will continue from the "
"included model.",
)
training_parser.add_argument(
"--trained-model",
dest="trained_model",
type=str,
help="Path to the trained model",
)
training_parser.add_argument(
"--model-weights",
dest="model_weights",
type=str,
help="Path to existing model weights",
)
training_parser.add_argument(
"--network-depth",
dest="network_depth",
type=valid_model_depth,
default="50",
help="Resnet depth (based on He et al. (2015)",
)
training_parser.add_argument(
"--batch-size",
dest="batch_size",
type=check_positive_int,
default=16,
help="Training batch size",
)
training_parser.add_argument(
"--epochs",
dest="epochs",
type=check_positive_int,
default=100,
help="Number of training epochs",
)
training_parser.add_argument(
"--test-fraction",
dest="test_fraction",
type=float,
default=0.1,
help="Fraction of training data to use for validation",
)
training_parser.add_argument(
"--learning-rate",
dest="learning_rate",
type=check_positive_float,
default=0.0001,
help="Learning rate for training the model",
)
training_parser.add_argument(
"--no-augment",
dest="no_augment",
action="store_true",
help="Don't apply data augmentation",
)
training_parser.add_argument(
"--save-weights",
dest="save_weights",
action="store_true",
help="Only store the model weights, and not the full model. Useful to "
"save storage space.",
)
training_parser.add_argument(
"--no-save-checkpoints",
dest="no_save_checkpoints",
action="store_true",
help="Store the model at intermediate points during training",
)
training_parser.add_argument(
"--tensorboard",
action="store_true",
help="Log to output_directory/tensorboard",
)
training_parser.add_argument(
"--save-progress",
dest="save_progress",
action="store_true",
help="Save training progress to a .csv file",
)
training_parser = misc_parse(training_parser)
training_parser = model_parser(training_parser)
training_parser = download_directory_parser(training_parser)
args = training_parser.parse_args()
return args