ECG_SEQ_LAB_NET

class torch_ecg.models.ECG_SEQ_LAB_NET(classes: Sequence[str], n_leads: int, config: CFG | None = None)

Bases: ECG_CRNN

SOTA model from CPSC2019 challenge.

Sequence labeling nets, for wave delineation, QRS complex detection, etc. Proposed in [Cai and Hu[1]].

pipeline

(multi-scopic, etc.) cnn –> head ((bidi-lstm –>) “attention” –> seq linear) -> output

Parameters:

• classes (List[str]) – List of the classes for sequence labeling.

• n_leads (int) – Number of leads (number of input channels).

• config (dict, optional) – Other hyper-parameters, including kernel sizes, etc. Refer to corresponding config file.

References

[1]

Wenjie Cai and Danqin Hu. QRS Complex Detection using Novel Deep Learning Neural Networks. IEEE Access, 8:97082–97089, 2020. doi:10.1109/access.2020.2997473.

compute_output_shape(seq_len: int | None = None, batch_size: int | None = None) → Sequence[int | None]

Compute the output shape of the model.

Parameters:

• seq_len (int, optional) – Length of the 1d input signal tensor.

• batch_size (int, optional) – Batch size of the input signal tensor.

Returns:

output_shape – The output shape of the model.

Return type:

sequence

forward(input: Tensor) → Tensor

Forward pass.

Parameters:

input (torch.Tensor) – Input tensor, of shape (batch_size, channels, seq_len).

Returns:

pred – Output tensor, of shape (batch_size, seq_len, n_classes)

Return type:

torch.Tensor

classmethod from_v1(v1_ckpt: str, device: device | None = None, return_config: bool = False) → ECG_SEQ_LAB_NET | Tuple[ECG_SEQ_LAB_NET, dict]

Convert the v1 model to the current version.

Parameters:

• v1_ckpt (str) – Path to the v1 checkpoint file.

• device (torch.device, optional) – The device to load the model to. Defaults to “cuda” if available, otherwise “cpu”.

• return_config (bool, default False) – Whether to return the config dict.

Returns:

• model (ECG_SEQ_LAB_NET) – The converted model.

• config (dict) – The config dict. (if return_config is True)