def __init__(self, aam, lk_algorithm_cls=WibergInverseCompositional,
n_shape=None, n_appearance=None, sampling=None):
self._model = aam
self._check_n_shape(n_shape)
self._check_n_appearance(n_appearance)
self._sampling = checks.check_sampling(sampling, aam.n_scales)
self._set_up(lk_algorithm_cls)
def __init__(self, images, aam, group=None, bounding_box_group_glob=None,
n_shape=None, n_appearance=None, sampling=None,
sd_algorithm_cls=ProjectOutNewton,
n_iterations=6, n_perturbations=30,
perturb_from_gt_bounding_box=noisy_shape_from_bounding_box,
batch_size=None, verbose=False):
self.aam = aam
# Check parameters
checks.set_models_components(aam.shape_models, n_shape)
checks.set_models_components(aam.appearance_models, n_appearance)
self._sampling = checks.check_sampling(sampling, aam.n_scales)
# patch_feature and patch_shape are not actually
# used because they are fully defined by the AAM already. Therefore,
# we just leave them as their 'defaults' because they won't be used.
super(SupervisedDescentAAMFitter, self).__init__(
images, group=group, bounding_box_group_glob=bounding_box_group_glob,
reference_shape=self.aam.reference_shape,
sd_algorithm_cls=sd_algorithm_cls,
holistic_features=self.aam.holistic_features,
diagonal=self.aam.diagonal,
scales=self.aam.scales, n_iterations=n_iterations,
n_perturbations=n_perturbations,
perturb_from_gt_bounding_box=perturb_from_gt_bounding_box,
batch_size=batch_size, verbose=verbose)
def __init__(self,
aps,
gn_algorithm_cls=Inverse,
n_shape=None,
weight=200.,
sampling=None):
# Check parameters
checks.set_models_components(aps.shape_models, n_shape)
self._sampling = checks.check_sampling(sampling, aps.n_scales)
self.weight = checks.check_multi_scale_param(aps.n_scales,
(float, int), 'weight',
weight)
# Get list of algorithm objects per scale
algorithms = []
for j in list(range(aps.n_scales)):
# create the interface object
interface = GaussNewtonBaseInterface(
appearance_model=aps.appearance_models[j],
deformation_model=aps.deformation_models[j],
transform=aps.shape_models[j],
weight=self.weight[j],
use_procrustes=aps.use_procrustes,
template=aps.appearance_models[j].mean(),
sampling=self._sampling[j],
patch_shape=aps.patch_shape[j],
patch_normalisation=aps.patch_normalisation[j])
# create the algorithm object and append it
algorithms.append(gn_algorithm_cls(interface))
# Call superclass
super(GaussNewtonAPSFitter, self).__init__(aps=aps,
algorithms=algorithms)
def __init__(self, aps, gn_algorithm_cls=Inverse, n_shape=None,
weight=200., sampling=None):
# Check parameters
checks.set_models_components(aps.shape_models, n_shape)
self._sampling = checks.check_sampling(sampling, aps.n_scales)
self.weight = checks.check_multi_scale_param(
aps.n_scales, (float, int), 'weight', weight)
# Get list of algorithm objects per scale
algorithms = []
for j in list(range(aps.n_scales)):
# create the interface object
interface = GaussNewtonBaseInterface(
appearance_model=aps.appearance_models[j],
deformation_model=aps.deformation_models[j],
transform=aps.shape_models[j], weight=self.weight[j],
use_procrustes=aps.use_procrustes,
template=aps.appearance_models[j].mean(),
sampling=self._sampling[j], patch_shape=aps.patch_shape[j],
patch_normalisation=aps.patch_normalisation[j])
# create the algorithm object and append it
algorithms.append(gn_algorithm_cls(interface))
# Call superclass
super(GaussNewtonAPSFitter, self).__init__(aps=aps,
algorithms=algorithms)
def __init__(self, atm, lk_algorithm_cls=InverseCompositional,
n_shape=None, sampling=None):
self._model = atm
# Check parameters
checks.set_models_components(atm.shape_models, n_shape)
self._sampling = checks.check_sampling(sampling, atm.n_scales)
# Set up algorithm
self._set_up(lk_algorithm_cls)
def __init__(self,
atm,
lk_algorithm_cls=InverseCompositional,
n_shape=None,
sampling=None):
self._model = atm
checks.set_models_components(atm.shape_models, n_shape)
self._sampling = checks.check_sampling(sampling, atm.n_scales)
self._set_up(lk_algorithm_cls)
def __init__(self,
aam,
lk_algorithm_cls=WibergInverseCompositional,
n_shape=None,
n_appearance=None,
sampling=None):
self._model = aam
self._check_n_shape(n_shape)
self._check_n_appearance(n_appearance)
self._sampling = checks.check_sampling(sampling, aam.n_scales)
self._set_up(lk_algorithm_cls)
def __init__(self, aam, lk_algorithm_cls=WibergInverseCompositional,
n_shape=None, n_appearance=None, sampling=None):
# Check parameters
checks.set_models_components(aam.shape_models, n_shape)
checks.set_models_components(aam.appearance_models, n_appearance)
self._sampling = checks.check_sampling(sampling, aam.n_scales)
# Get list of algorithm objects per scale
interfaces = aam.build_fitter_interfaces(self._sampling)
algorithms = [lk_algorithm_cls(interface) for interface in interfaces]
# Call superclass
super(LucasKanadeAAMFitter, self).__init__(aam=aam,
algorithms=algorithms)
def __init__(self, atm, lk_algorithm_cls=InverseCompositional,
n_shape=None, sampling=None):
# Store model
self._model = atm
# Check parameters
checks.set_models_components(atm.shape_models, n_shape)
self._sampling = checks.check_sampling(sampling, atm.n_scales)
# Get list of algorithm objects per scale
interfaces = atm.build_fitter_interfaces(self._sampling)
algorithms = [lk_algorithm_cls(interface) for interface in interfaces]
# Call superclass
super(LucasKanadeATMFitter, self).__init__(
scales=atm.scales, reference_shape=atm.reference_shape,
holistic_features=atm.holistic_features, algorithms=algorithms)
def __init__(self,
aam,
lk_algorithm_cls=WibergInverseCompositional,
n_shape=None,
n_appearance=None,
sampling=None):
# Check parameters
checks.set_models_components(aam.shape_models, n_shape)
checks.set_models_components(aam.appearance_models, n_appearance)
self._sampling = checks.check_sampling(sampling, aam.n_scales)
# Get list of algorithm objects per scale
interfaces = aam.build_fitter_interfaces(self._sampling)
algorithms = [lk_algorithm_cls(interface) for interface in interfaces]
# Call superclass
super(LucasKanadeAAMFitter, self).__init__(aam=aam,
algorithms=algorithms)
def __init__(self,
atm,
lk_algorithm_cls=InverseCompositional,
n_shape=None,
sampling=None):
# Store model
self._model = atm
# Check parameters
checks.set_models_components(atm.shape_models, n_shape)
self._sampling = checks.check_sampling(sampling, atm.n_scales)
# Get list of algorithm objects per scale
interfaces = atm.build_fitter_interfaces(self._sampling)
algorithms = [lk_algorithm_cls(interface) for interface in interfaces]
# Call superclass
super(LucasKanadeATMFitter,
self).__init__(scales=atm.scales,
reference_shape=atm.reference_shape,
holistic_features=atm.holistic_features,
algorithms=algorithms)
def __init__(self,
images,
aam,
group=None,
bounding_box_group_glob=None,
n_shape=None,
n_appearance=None,
sampling=None,
sd_algorithm_cls=ProjectOutNewton,
n_iterations=6,
n_perturbations=30,
perturb_from_gt_bounding_box=noisy_shape_from_bounding_box,
batch_size=None,
verbose=False):
self.aam = aam
# Check parameters
checks.set_models_components(aam.shape_models, n_shape)
checks.set_models_components(aam.appearance_models, n_appearance)
self._sampling = checks.check_sampling(sampling, aam.n_scales)
# patch_feature and patch_shape are not actually
# used because they are fully defined by the AAM already. Therefore,
# we just leave them as their 'defaults' because they won't be used.
super(SupervisedDescentAAMFitter, self).__init__(
images,
group=group,
bounding_box_group_glob=bounding_box_group_glob,
reference_shape=self.aam.reference_shape,
sd_algorithm_cls=sd_algorithm_cls,
holistic_features=self.aam.holistic_features,
diagonal=self.aam.diagonal,
scales=self.aam.scales,
n_iterations=n_iterations,
n_perturbations=n_perturbations,
perturb_from_gt_bounding_box=perturb_from_gt_bounding_box,
batch_size=batch_size,
verbose=verbose)
def __init__(self, unified_aam_clm,
algorithm_cls=AlternatingRegularisedLandmarkMeanShift,
n_shape=None, n_appearance=None, sampling=None):
self._model = unified_aam_clm
# Check parameters
checks.set_models_components(self._model.shape_models, n_shape)
checks.set_models_components(self._model.appearance_models,
n_appearance)
self._sampling = checks.check_sampling(sampling, self._model.n_scales)
# Get list of algorithm objects per scale
interfaces = unified_aam_clm.build_fitter_interfaces(self._sampling)
algorithms = [algorithm_cls(interface,
self._model.expert_ensembles[k],
self._model.patch_shape[k],
self._model.response_covariance)
for k, interface in enumerate(interfaces)]
# Call superclass
super(UnifiedAAMCLMFitter, self).__init__(
scales=self._model.scales,
reference_shape=self._model.reference_shape,
holistic_features=self._model.holistic_features,
algorithms=algorithms)
def __init__(self, atm, algorithm_cls=InverseCompositional,
n_shape=None, sampling=None):
self._model = atm
checks.set_models_components(atm.shape_models, n_shape)
self._sampling = checks.check_sampling(sampling, atm.n_scales)
self._set_up(algorithm_cls)
