HHARDataset#

class torchdyno.data.datasets.hhar.HHARDataset[source]#

Bases: Dataset

HHAR dataset.

The HHAR dataset is a dataset of human activity recognition (HAR) collected from smartphones and smartwatches. The dataset contains accelerometer and gyroscope data from different devices and users. The dataset is used to classify the activity of the user based on the sensor data.

Methods

__init__(root, user, context[, seq_length])

Initialize the HHAR dataset.

preprocess()

Attributes

CONTEXTS

USERS

processed_path

raw_path

seq_length

Return the sequence length.
__init__(root, user, context, seq_length=200)[source]#

Initialize the HHAR dataset.

Parameters:

  • root (str) – Root directory of the dataset.

  • user (str) – User to load.

  • context (int) – Context to load.

  • seq_length (Optional[int], optional) – Sequence length. If None, the dataset is not split into sequences. Defaults to 200.

property seq_length#

Return the sequence length.