PreprocManager¶

class torch_ecg.preprocessors.PreprocManager(*pps: Tuple[Module, ...] | None, random: bool = False, inplace: bool = True)[source]¶

Bases: ReprMixin, Module

Manager class for preprocessors.

Parameters:

pps (Tuple[torch.nn.Module], optional) – The sequence of preprocessors to be added to the manager.
random (bool, default False) – Whether to apply the preprocessors in random order.
inplace (bool, default True) – Whether to apply the preprocessors in-place.

Examples

import torch
from torch_ecg.cfg import CFG
from torch_ecg.preprocessors import PreprocManager

config = CFG(
    random=False,
    bandpass={"fs":500},
    normalize={"method": "min-max"},
)
ppm = PreprocManager.from_config(config)
sig = torch.randn(2, 12, 8000)
sig = ppm(sig)

add_(pp: Module, pos: int = -1) → None[source]¶

Add a (custom) preprocessor to the manager.

This method is preferred against directly manipulating the internal list of preprocessors via PreprocManager.preprocessors.append(pp).

Parameters:

pp (torch.nn.Module) – The preprocessor to be added.
pos (int, default -1) – The position to insert the preprocessor. Should be >= -1, with -1 being the indicator of the end.

extra_repr_keys() → List[str][source]¶: Extra keys for __repr__() and __str__().

forward(sig: Tensor) → Tensor[source]¶

Apply the preprocessors to the signal tensor.

Parameters:: sig (torch.Tensor) – The signal tensor to be preprocessed.
Returns:: The preprocessed signal tensor.
Return type:: torch.Tensor

classmethod from_config(config: dict) → PreprocManager[source]¶

Initialize a PreprocManager instance from a configuration.

Parameters:: config (dict) – The configuration of the preprocessors, better to be an OrderedDict.
Returns:: ppm – A new instance of PreprocManager.
Return type:: PreprocManager

rearrange(new_ordering: List[str]) → None[source]¶

Rearrange the preprocessors.

Parameters:: new_ordering (List[str]) – The new ordering of the preprocessors.
Return type:: None