anomalib.utils.cli

Anomalib CLI.

Submodules

Package Contents

Classes

AnomalibCLI

Implementation of a fully configurable CLI tool for anomalib.
class anomalib.utils.cli.AnomalibCLI(model_class: Optional[Union[Type[pytorch_lightning.LightningModule], Callable[Ellipsis, pytorch_lightning.LightningModule]]] = None, datamodule_class: Optional[Union[Type[pytorch_lightning.LightningDataModule], Callable[Ellipsis, pytorch_lightning.LightningDataModule]]] = None, save_config_callback: Optional[Type[pytorch_lightning.utilities.cli.SaveConfigCallback]] = SaveConfigCallback, save_config_filename: str = 'config.yaml', save_config_overwrite: bool = False, save_config_multifile: bool = False, trainer_class: Union[Type[pytorch_lightning.Trainer], Callable[Ellipsis, pytorch_lightning.Trainer]] = Trainer, trainer_defaults: Optional[Dict[str, Any]] = None, seed_everything_default: Optional[int] = None, description: str = 'Anomalib trainer command line tool', env_prefix: str = 'Anomalib', env_parse: bool = False, parser_kwargs: Optional[Union[Dict[str, Any], Dict[str, Dict[str, Any]]]] = None, subclass_mode_model: bool = False, subclass_mode_data: bool = False, run: bool = True, auto_registry: bool = True)[source]

Bases: pytorch_lightning.utilities.cli.LightningCLI

Implementation of a fully configurable CLI tool for anomalib.

The advantage of this tool is its flexibility to configure the pipeline from both the CLI and a configuration file (.yaml or .json). It is even possible to use both the CLI and a configuration file simultaneously. For more details, the reader could refer to PyTorch Lightning CLI documentation.

add_arguments_to_parser(parser: pytorch_lightning.utilities.cli.LightningArgumentParser) → None

Add default arguments.

Parameters

parser (LightningArgumentParser) – Lightning Argument Parser.

__set_default_root_dir() → None

Sets the default root directory depending on the subcommand type. <train, fit, predict, tune.>.

__set_callbacks() → None

Sets the default callbacks used within the pipeline.

before_instantiate_classes() → None

Modify the configuration to properly instantiate classes.