EMGZarrDataset#

class myoverse.datasets.loader.EMGZarrDataset(zarr_file, subset_name, target_name, emg_dtype=<class 'numpy.float32'>, target_dtype=<class 'numpy.float32'>, sampling_frequency=2048.0, input_data_class=<class 'myoverse.datatypes.EMGData'>, target_data_class=<class 'myoverse.datatypes.KinematicsData'>, input_augmentation_pipeline=None, input_augmentation_probabilities=None, target_augmentation_pipeline=None, target_augmentation_probabilities=None, cache_size=100)[source]#

Dataset class for loading EMG data from Zarr files.

Parameters:

  • zarr_file (Path) – The path to the zarr file

  • subset_name (str) – The name of the subset to load (e.g., “training”, “validation”, “testing”)

  • target_name (str) – The name of the target data

  • emg_dtype (np.dtype, optional) – The data type of the EMG data, by default np.float32

  • target_dtype (np.dtype, optional) – The data type of the target data, by default np.float32

  • sampling_frequency (float, optional) – The sampling frequency of the EMG data in Hz, by default 2048.0

  • input_data_class (Type[_Data], optional) – The class to use for input data, by default EMGData

  • target_data_class (Type[_Data], optional) – The class to use for target data, by default KinematicsData

  • input_augmentation_pipeline (list[list[FilterBaseClass]], optional) – The augmentation pipeline for the input data

  • input_augmentation_probabilities (Sequence[float], optional) – The probabilities for each input augmentation pipeline

  • target_augmentation_pipeline (list[list[FilterBaseClass]], optional) – The augmentation pipeline for the target data

  • target_augmentation_probabilities (Sequence[float], optional) – The probabilities for each target augmentation pipeline

  • cache_size (int, optional) – The maximum number of items to cache, by default 100

Methods

__getitem__(idx)

__init__(zarr_file, subset_name, target_name)

__len__()

_validate_augmentation_probabilities()

Validate that augmentation probabilities are valid.
On this page

This Page