pose_format.pose_representation module

Classes:

PoseRepresentation(header[, rep_modules1, ...])

Represents a pose using various representation modules.

class pose_format.pose_representation.PoseRepresentation(header, rep_modules1=[], rep_modules2=[], rep_modules3=[])[source]

Bases: object

Represents a pose using various representation modules.

Parameters:

header (PoseHeader) – Header information about the pose.
rep_modules1 (List, optional) – List of modules that use a point-based representation. Defaults to an empty list.
rep_modules2 (List, optional) – List of modules that use a limb-based representation. Defaults to an empty list.
rep_modules3 (List, optional) – List of modules that use a triangle-based representation. Defaults to an empty list.

input_size

Combined point count across all components of the header.

Type:: int

rep_modules1_size

The size represented by the point-based modules.

Type:: int

rep_modules2_size

The size represented by the limb-based modules.

Type:: int

rep_modules3_size

The size represented by the triangle-based modules.

Type:: int

output_size

The overall size of the output representation.

Type:: int

Methods:

`calc_output_size`()	Calculate total size of output representation, based on active modules
`get_limbs_points`()	Get points that define limbs
`get_points`(tensor, points)	get points from a given tensor
`get_triangles_points`()	Get points that make up triangles.
`group_embeds`(embeds)	Groups given embeddings into a desired format.
`permute`(src, shape)	Permutes given tensor according to shape.

calc_output_size()[source]

Calculate total size of output representation, based on active modules

Returns:: Total size of the module representation.
Return type:: int

get_limbs_points()[source]

Get points that define limbs

Returns:: Two lists containing points that define the start and end of each limb.
Return type:: Tuple[List, List]

get_points(tensor, points)[source]

get points from a given tensor

Parameters:

tensor (torch.Tensor) – Tensor from which you need the points.
points (List[int]) – Indices of points that need to be extracted.

Returns:

Gotten points from tensor.

Return type:

torch.Tensor

get_triangles_points()[source]

Get points that make up triangles.

Returns:: Three lists which have points that define each corner of the triangles.
Return type:: Tuple[List, List, List]

group_embeds(embeds)[source]

Groups given embeddings into a desired format. Must be implemented by subclasses.

Parameters:: embeds (List) – List of tensor embeddings, tensor size: (embed_size, Batch, Len).
Raises:: NotImplementedError – If the method is not implemented by subclasses.
Return type:: Size (Batch, Len, embed_size)

permute(src, shape)[source]

Permutes given tensor according to shape.

Parameters:

src (torch.Tensor) – Tensor to permute.
shape (tuple) – Desired shape of permuted tensor.

Raises:

NotImplementedError – If method is not implemented by subclasses.