The PyTorch-FWD package

The PyTorch-FWD package#

pytorchfwd.fwd module#

Frechet Wavelet Distance computation.

pytorchfwd.fwd.calculate_path_statistics(path: str, wavelet: str, max_level: int, log_scale: bool, batch_size: int) → Tuple[ndarray, ...][source]#

Compute mean and sigma for given path.

Parameters:

  • path (str) – npz path or image directory.

  • wavelet (str) – Choice of wavelet.

  • max_level (int) – Decomposition level.

  • log_scale (bool) – Apply log scale.

  • batch_size (int) – Batch size for packet decomposition.

Raises:

ValueError – Error if mu and sigma cannot be calculated.

Returns:

Tuple containing mean and sigma for each packet.

Return type:

Tuple[np.ndarray, …]

pytorchfwd.fwd.compute_fwd(paths: List[str], wavelet: str, max_level: int, log_scale: bool, batch_size: int) → float[source]#

Compute Frechet Wavelet Distance.

Parameters:

  • paths (List[str]) – List containing path of source and generated images.

  • wavelet (str) – Choice of wavelet.

  • max_level (int) – Decomposition level.

  • log_scale (bool) – Apply log scale.

  • batch_size (int) – Batch size for packet decomposition.

Raises:

RuntimeError – Error if path doesn’t exist.

Returns:

Frechet Wavelet Distance.

Return type:

float

pytorchfwd.fwd.compute_packet_statistics(dataloader: DataLoader, wavelet: str, max_level: int, log_scale: bool) → Tuple[ndarray, ...][source]#

Compute wavelet packet transform across batches.

Parameters:

  • dataloader (th.utils.data.DataLoader) – Torch dataloader.

  • wavelet (str) – Choice of wavelet.

  • max_level (int) – Wavelet decomposition level.

  • log_scale (bool) – Apply log scale.

Returns:

Mean and sigma for each packet.

Return type:

Tuple[np.ndarray, …]

pytorchfwd.fwd.main()[source]#

Compute FWD given paths.

Contents