WaveformInput

class torch_ecg.components.WaveformInput(config: InputConfig)

Bases: BaseInput

Waveform input.

Examples

>>> from torch_ecg.cfg import DEFAULTS
>>> BATCH_SIZE = 32
>>> N_CHANNELS = 12
>>> N_SAMPLES = 5000
>>> input_config = InputConfig(
...     input_type="waveform",
...     n_channels=N_CHANNELS,
...     n_samples=N_SAMPLES,
... )
>>> inputer = WaveformInput(input_config)
>>> waveform = torch.randn(BATCH_SIZE, N_CHANNELS, N_SAMPLES)
>>> inputer(waveform).shape
torch.Size([32, 12, 5000])
>>> waveform = DEFAULTS.RNG.uniform(size=(N_CHANNELS, N_SAMPLES))
>>> inputer(waveform).shape
torch.Size([1, 12, 5000])
>>> input_config = InputConfig(
...     input_type="waveform",
...     n_channels=N_CHANNELS,
...     n_samples=N_SAMPLES,
...     ensure_batch_dim=False,
... )
>>> inputer = WaveformInput(input_config)
>>> waveform = DEFAULTS.RNG.uniform(size=(N_CHANNELS, N_SAMPLES))
>>> inputer(waveform).shape
torch.Size([12, 5000])

from_waveform(waveform: ndarray | Tensor) → Tensor

Converts the input ndarray or Tensor waveform to a Tensor.

Parameters:

waveform (numpy.ndarray or torch.Tensor) – The waveform to be transformed, of shape (batch_size, n_channels, n_samples) or (n_channels, n_samples).

Returns:

The transformed waveform, of shape (batch_size, n_channels, n_samples).

Return type:

torch.Tensor

Note

If the input is a 2D tensor, then the batch dimension is added (batch_size = 1).