PCALoss

class lightning_pose.losses.losses.PCALoss

Bases: Loss

Penalize predictions that fall outside a low-dimensional subspace.

Methods Summary

__call__(keypoints_pred[, stage])	Call self as a function.
compute_loss(predictions)
remove_nans(**kwargs)

Methods Documentation

__call__(keypoints_pred: Tensor, stage: Literal['train', 'val', 'test'] | None = None, **kwargs) → Tensor, {'__torchtyping__': True, 'details': ((),), 'cls_name': 'TensorType'}], list[dict]]

Call self as a function.

compute_loss(predictions: Tensor, {'__torchtyping__': True, 'details': ('num_samples', 'sample_dim'), 'cls_name': 'TensorType'}]) → Tensor, {'__torchtyping__': True, 'details': ('num_samples', -1,), 'cls_name': 'TensorType'}]

remove_nans(**kwargs)

__init__(loss_name: Literal['pca_singleview', 'pca_multiview'], components_to_keep: int | float = 0.95, empirical_epsilon_percentile: float = 0.99, epsilon: float | None = None, empirical_epsilon_multiplier: float = 1.0, mirrored_column_matches: ListConfig | list | None = None, columns_for_singleview_pca: ListConfig | list | None = None, data_module: BaseDataModule | UnlabeledDataModule | None = None, log_weight: float = 0.0, device: Literal['cpu', 'cuda'] | device = 'cpu', centering_method: Literal['mean', 'median'] | None = None, **kwargs) → None

Parameters:

• data_module – give losses access to data for computing data-specific loss params

• epsilon – loss values below epsilon will be zeroed out

• log_weight – natural log of the weight in front of the loss term in the final objective function

__new__(**kwargs)