sleap.gui.learning.scopedkeydict

Conversion between flat (form data) and hierarchical (config object) dicts.

class sleap.gui.learning.scopedkeydict.ScopedKeyDict(key_val_dict: Dict[str, Any])

Class to support conversion between flat and hierarchical dictionaries.

Flat dictionaries have scoped keys, e.g., “foo.bar”. These typically come from user-editable forms.

Hierarchical dictionaries have dictionaries as values of other dictionaries, e.g., {“foo”: {“bar”: … } }. These are typically used when serializing and deserializing objects.

key_val_dict

Data stores in a flat dictionary with scoped keys.

classmethod from_hierarchical_dict(hierarch_dict: dict)

Constructs object (with flat dict) from hierarchical dictionary.

classmethod set_hierarchical_key_val(current_dict: dict, key: str, val: Any)

Sets value in hierarchical dict via scoped key.

to_hierarchical_dict() → dict

Converts internal flat dictionary to hierarchical dictionary.

sleap.gui.learning.scopedkeydict.apply_cfg_transforms_to_key_val_dict(key_val_dict: dict)

Transforms data from form to correct data types before converting to object.

Parameters

key_val_dict – Flat dictionary from TrainingEditorWidget.

Returns

None, modifies dict in place.

sleap.gui.learning.scopedkeydict.find_backbone_name_from_key_val_dict(key_val_dict: dict)

Find the backbone model name from the config dictionary.

sleap.gui.learning.scopedkeydict.make_model_config_from_key_val_dict(key_val_dict: dict) → sleap.nn.config.model.ModelConfig

Make ModelConfig object from flat dictionary.

Parameters

key_val_dict – Flat dictionary from TrainingEditorWidget.

Returns

The ModelConfig object.

sleap.gui.learning.scopedkeydict.make_training_config_from_key_val_dict(key_val_dict: dict) → sleap.nn.config.training_job.TrainingJobConfig

Make TrainingJobConfig object from flat dictionary.

Parameters

key_val_dict – Flat dictionary from TrainingEditorWidget.

Returns

The TrainingJobConfig object.

sleap.gui.learning.scopedkeydict.resolve_strides_from_key_val_dict(key_val_dict: dict, backbone_name: str) → Tuple[int, int]

Find the valid max and output strides from the config dictionary.