training_signal

@evented
@dataclass
class TrainingSignal:
    """Training signal class.

    This class holds the parameters required to run the training thread. These
    parameters should be set whenever the user interact with the corresponding UI
    elements. An instance of the class is then passed to the training worker.
    """

    if TYPE_CHECKING:
        events: TrainingSignalGroup
        """Attribute allowing the registration of parameter-specific listeners."""

    # signals used to change states across widgets
    algorithm: str = "n2v"
    """Algorithm used for training."""

    use_channels: bool = False
    """Whether the data has channels."""

    is_3d: bool = False
    """Whether the data is 3D."""

    # parameters set by widgets for training
    work_dir: Path = HOME
    """Directory where the checkpoints and logs are saved."""

    load_from_disk: bool = True
    """Whether to load the images from disk or from the viewer."""

    if _has_napari:
        layer_train: Image = None
        """Layer containing the training data."""

        layer_train_target: Image = None
        """Layer containing the training target data."""

        layer_val: Image = None
        """Layer containing the validation data."""

        layer_val_target: Image = None
        """Layer containing the validation target data."""

    path_train: str = ""
    """Path to the training data."""

    path_train_target: str = ""
    """Path to the training target data."""

    path_val: str = ""
    """Path to the validation data."""

    path_val_target: str = ""
    """Path to the validation target."""

    axes: str = "YX"
    """Axes of the data."""

    patch_size_xy: int = 64
    """Size of the patches along the X and Y dimensions."""

    patch_size_z: int = 16
    """Size of the patches along the Z dimension."""

    n_epochs: int = 30
    """Number of epochs."""

    batch_size: int = 16
    """Batch size."""

    experiment_name = ""
    """Name of the experiment, used to export the model and save checkpoints."""

    x_flip: bool = True
    """Whether to apply flipping along the X dimension during augmentation."""

    y_flip: bool = True
    """Whether to apply flipping along the Y dimension during augmentation."""

    rotations: bool = True
    """Whether to apply rotations during augmentation."""

    independent_channels: bool = True
    """Whether to train the channels independently."""

    n_channels_n2v: int = 1
    """Number of channels when training Noise2Void."""

    n_channels_in_care: int = 1
    """Number of input channels when training CARE and Noise2Noise."""

    n_channels_out_care: int = 1
    """Number of output channels when training CARE and Noise2Noise."""

    use_n2v2: bool = False
    """Whether to use N2V2."""

    depth: int = 2
    """Depth of the U-Net."""

    num_conv_filters: int = 32
    """Number of convolutional filters in the first layer."""

    val_percentage: float = 0.1
    """Percentage of the training data used for validation."""

    val_minimum_split: int = 1
    """Minimum number of patches or images in the validation set."""

class TrainingSignalGroup(SignalGroup):
    """Signal group for the training status dataclass."""

    # only parameters that have observers are listed here
    algorithm: SignalInstance
    """Algorithm used for training."""

    use_channels: SignalInstance
    """Whether tthe data has channels."""

    is_3d: SignalInstance
    """Whether the data is 3D."""