poseutils.datasets.processed package

Submodules

poseutils.datasets.processed.TransformedDataset module

class poseutils.datasets.processed.TransformedDataset.TransformedDataset(dataset=None, transformations2d=[], transformations3d=[], njnts=14)

Bases: object

This class takes a unprocessed dataset and applies different transformations to it.

Parameters

• dataset (poseutils.datasets.unprocessed.Dataset, optional) – Unprocessed dataset, defaults to None

• transformations2d (list(poseutils.datasets.transformation.Transformation), defaults to []) – List of transformations to apply on 2d data

• transformations3d (list(poseutils.datasets.transformation.Transformation), optional) – List of transformations to apply on 3d data, defaults to []

• njnts (int, optional) – Number of joints i.e. joint configuration (14 or 16), defaults to 14

apply2d(transformations)

Applies transformations to the 2d dataset (both training and validation)

Parameters

transformations (list(poseutils.datasets.transformation.Transformation)) – List of transformations to apply on 2d data

apply3d(transformations)

Applies transformations to the 3d dataset (both training and validation)

Parameters

transformations (list(poseutils.datasets.transformation.Transformation)) – List of transformations to apply on 3d data

calculate_metrics()

Calculates mean and standard devation from the training dataset (2d and 3d)

get_2d_train()

Returns 2d training data

Returns

Joint positions 2d (NxMx2), M = 14 or 16

Return type

numpy.ndarray

get_2d_valid()

Returns 2d validation data

Returns

Joint positions 2d (NxMx2), M = 14 or 16

Return type

numpy.ndarray

get_3d_train()

Returns 3d training data

Returns

Joint positions 3d (NxMx3), M = 14 or 16

Return type

numpy.ndarray

get_3d_valid()

Returns 3d validation data

Returns

Joint positions 3d (NxMx3), M = 14 or 16

Return type

numpy.ndarray

set_train(d2d, d3d)

Set training data

Parameters

• d2d (numpy.ndarray) – 2d data to set (NxMx2), M = 14 or 16

• d3d (numpy.ndarray) – 3d data to set (NxMx3), M = 14 or 16

set_valid(d2d, d3d)

Set validation data

Parameters

• d2d (numpy.ndarray) – 2d data to set (NxMx2), M = 14 or 16

• d3d (numpy.ndarray) – 3d data to set (NxMx3), M = 14 or 16

transform()

Module contents