def __init__(self, data, **kwargs):
from menpofit.transform import DifferentiableAlignmentSimilarity
aligned_shapes = align_shapes(data)
self.mean = mean_pointcloud(aligned_shapes)
# Default target is the mean
self._target = self.mean
self.transform = DifferentiableAlignmentSimilarity(
self.target, self.target)
def __init__(self, data, **kwargs):
from menpofit.transform import DifferentiableAlignmentSimilarity
aligned_shapes = align_shapes(data)
self.mean = mean_pointcloud(aligned_shapes)
# Default target is the mean
self._target = self.mean
self.transform = DifferentiableAlignmentSimilarity(self.target,
self.target)
class GlobalSimilarityModel(Targetable, Vectorizable):
r"""
Class for creating a model that represents a global similarity transform
(in-plane rotation, scaling, translation).
Parameters
----------
data : `list` of `menpo.shape.PointCloud`
The `list` of shapes to use as training data.
"""
def __init__(self, data, **kwargs):
from menpofit.transform import DifferentiableAlignmentSimilarity
aligned_shapes = align_shapes(data)
self.mean = mean_pointcloud(aligned_shapes)
# Default target is the mean
self._target = self.mean
self.transform = DifferentiableAlignmentSimilarity(
self.target, self.target)
@property
def n_weights(self):
r"""
The number of parameters in the linear model.
:type: `int`
"""
return 4
@property
def weights(self):
r"""
The weights of the model.
:type: ``(n_weights,)`` `ndarray`
"""
return self.transform.as_vector()
@property
def target(self):
r"""
The current `menpo.shape.PointCloud` that this object produces.
:type: `menpo.shape.PointCloud`
"""
return self._target
def set_target(self, new_target):
r"""
Update this object so that it attempts to recreate the ``new_target``.
Parameters
----------
new_target : `menpo.shape.PointCloud`
The new target that this object should try and regenerate.
"""
self.transform.set_target(new_target)
self._target = self.transform.apply(self.mean)
return self
def _as_vector(self):
r"""
Return the current parameters of this transform - this is the
just the linear model's weights
Returns
-------
params : (`n_parameters`,) ndarray
The vector of parameters
"""
return self.transform.as_vector()
def _from_vector_inplace(self, vector):
self.transform._from_vector_inplace(vector)
self._target = self.transform.apply(self.mean)
@property
def n_dims(self):
r"""
The number of dimensions of the spatial instance of the model.
:type: `int`
"""
return self.mean.n_dims
def d_dp(self, _):
r"""
Returns the Jacobian of the similarity model reshaped in order to have
the standard Jacobian shape, i.e. ``(n_points, n_weights, n_dims)``
which maps to ``(n_features, n_components, n_dims)`` on the linear model.
Returns
-------
jacobian : ``(n_features, n_components, n_dims)`` `ndarray`
The Jacobian of the model in the standard Jacobian shape.
"""
# Always evaluated at the mean shape
return self.transform.d_dp(self.mean.points)
class GlobalSimilarityModel(Targetable, Vectorizable):
def __init__(self, data, **kwargs):
from menpofit.transform import DifferentiableAlignmentSimilarity
aligned_shapes = align_shapes(data)
self.mean = mean_pointcloud(aligned_shapes)
# Default target is the mean
self._target = self.mean
self.transform = DifferentiableAlignmentSimilarity(
self.target, self.target)
@property
def n_weights(self):
r"""
The number of parameters in the linear model.
:type: int
"""
return 4
@property
def weights(self):
r"""
In this simple :map:`ModelInstance` the weights are just the weights
of the model.
"""
return self.transform.as_vector()
@property
def target(self):
return self._target
def set_target(self, new_target):
self.transform.set_target(new_target)
self._target = self.transform.apply(self.mean)
return self
def _as_vector(self):
r"""
Return the current parameters of this transform - this is the
just the linear model's weights
Returns
-------
params : (`n_parameters`,) ndarray
The vector of parameters
"""
return self.transform.as_vector()
def _from_vector_inplace(self, vector):
self.transform._from_vector_inplace(vector)
self._target = self.transform.apply(self.mean)
@property
def n_dims(self):
r"""
The number of dimensions of the spatial instance of the model
:type: int
"""
return self.mean.n_dims
def d_dp(self, _):
"""
Returns the Jacobian of the similarity model reshaped to have the
standard Jacobian shape:
n_points x n_params x n_dims
which maps to
n_features x n_components x n_dims
on the linear model
Returns
-------
jacobian : (n_features, n_components, n_dims) ndarray
The Jacobian of the model in the standard Jacobian shape.
"""
# Always evaluated at the mean shape
return self.transform.d_dp(self.mean.points)
class GlobalSimilarityModel(Targetable, Vectorizable):
def __init__(self, data, **kwargs):
from menpofit.transform import DifferentiableAlignmentSimilarity
aligned_shapes = align_shapes(data)
self.mean = mean_pointcloud(aligned_shapes)
# Default target is the mean
self._target = self.mean
self.transform = DifferentiableAlignmentSimilarity(self.target,
self.target)
@property
def n_weights(self):
r"""
The number of parameters in the linear model.
:type: int
"""
return 4
@property
def weights(self):
r"""
In this simple :map:`ModelInstance` the weights are just the weights
of the model.
"""
return self.transform.as_vector()
@property
def target(self):
return self._target
def set_target(self, new_target):
self.transform.set_target(new_target)
self._target = self.transform.apply(self.mean)
return self
def _as_vector(self):
r"""
Return the current parameters of this transform - this is the
just the linear model's weights
Returns
-------
params : (`n_parameters`,) ndarray
The vector of parameters
"""
return self.transform.as_vector()
def _from_vector_inplace(self, vector):
self.transform._from_vector_inplace(vector)
self._target = self.transform.apply(self.mean)
@property
def n_dims(self):
r"""
The number of dimensions of the spatial instance of the model
:type: int
"""
return self.mean.n_dims
def d_dp(self, _):
"""
Returns the Jacobian of the similarity model reshaped to have the
standard Jacobian shape:
n_points x n_params x n_dims
which maps to
n_features x n_components x n_dims
on the linear model
Returns
-------
jacobian : (n_features, n_components, n_dims) ndarray
The Jacobian of the model in the standard Jacobian shape.
"""
# Always evaluated at the mean shape
return self.transform.d_dp(self.mean.points)
class GlobalSimilarityModel(Targetable, Vectorizable):
r"""
Class for creating a model that represents a global similarity transform
(in-plane rotation, scaling, translation).
Parameters
----------
data : `list` of `menpo.shape.PointCloud`
The `list` of shapes to use as training data.
"""
def __init__(self, data, **kwargs):
from menpofit.transform import DifferentiableAlignmentSimilarity
aligned_shapes = align_shapes(data)
self.mean = mean_pointcloud(aligned_shapes)
# Default target is the mean
self._target = self.mean
self.transform = DifferentiableAlignmentSimilarity(self.target,
self.target)
@property
def n_weights(self):
r"""
The number of parameters in the linear model.
:type: `int`
"""
return 4
@property
def weights(self):
r"""
The weights of the model.
:type: ``(n_weights,)`` `ndarray`
"""
return self.transform.as_vector()
@property
def target(self):
r"""
The current `menpo.shape.PointCloud` that this object produces.
:type: `menpo.shape.PointCloud`
"""
return self._target
def set_target(self, new_target):
r"""
Update this object so that it attempts to recreate the ``new_target``.
Parameters
----------
new_target : `menpo.shape.PointCloud`
The new target that this object should try and regenerate.
"""
self.transform.set_target(new_target)
self._target = self.transform.apply(self.mean)
return self
def _as_vector(self):
r"""
Return the current parameters of this transform - this is the
just the linear model's weights
Returns
-------
params : (`n_parameters`,) ndarray
The vector of parameters
"""
return self.transform.as_vector()
def _from_vector_inplace(self, vector):
self.transform._from_vector_inplace(vector)
self._target = self.transform.apply(self.mean)
@property
def n_dims(self):
r"""
The number of dimensions of the spatial instance of the model.
:type: `int`
"""
return self.mean.n_dims
def d_dp(self, _):
r"""
Returns the Jacobian of the similarity model reshaped in order to have
the standard Jacobian shape, i.e. ``(n_points, n_weights, n_dims)``
which maps to ``(n_features, n_components, n_dims)`` on the linear model.
Returns
-------
jacobian : ``(n_features, n_components, n_dims)`` `ndarray`
The Jacobian of the model in the standard Jacobian shape.
"""
# Always evaluated at the mean shape
return self.transform.d_dp(self.mean.points)
