def _build_shape_model(self, shapes, shape_graph, max_shape_components,
verbose=False):
# if the provided graph is None, then apply PCA, else use the GMRF
if shape_graph is not None:
pca_model = GMRFModel(
shapes, shape_graph, mode='concatenation', n_components=None,
dtype=np.float64, sparse=False, incremental=self.is_incremental,
verbose=verbose).principal_components_analysis()
return OrthoPDM(pca_model, max_n_components=max_shape_components)
else:
return OrthoPDM(shapes, max_n_components=max_shape_components)
def PointDistributionModel(imgFolder):
'''LOAD IMAGES'''
path_to_lfpw = Path(imgFolder)
training_shapes = []
for lg in print_progress(
mio.import_landmark_files(path_to_lfpw / '*.pts', verbose=True)):
training_shapes.append(lg['all'])
'''TRAIN PDM MODEL'''
shape_model = OrthoPDM(training_shapes, max_n_components=None)
'''MODIFY PARAMETERS'''
shape_model.n_active_components = 20
shape_model.n_active_components = 0.95
return shape_model
def set_target(self, target):
if target.n_points < self.target.n_points:
if target.n_points < self.n_landmarks:
target = PointCloud(target.points[:self.n_align_lms])
target = np.dot(np.dot(target.as_vector(), self.pinv_va), self.W)
else:
target = PointCloud(target.points[:self.n_landmarks])
target = np.dot(np.dot(target.as_vector(), self.pinv_v), self.W)
target = PointCloud(np.reshape(target, (-1, self.n_dims)))
OrthoPDM.set_target(self, target)
def set_target(self, 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.
"""
if target.n_points == self.n_landmarks:
# densify target
target = np.dot(np.dot(target.as_vector(), self.pinv_V), self.W)
target = PointCloud(np.reshape(target, (-1, self.n_dims)))
OrthoPDM.set_target(self, target)
def PDMModel(path, max_components=None):
training_shapes = []
for lg in print_progress(
mio.import_landmark_files(path / '*.pts', verbose=True)):
training_shapes.append(lg['all'])
# train source PDM model
shape_model = OrthoPDM(training_shapes, max_n_components=max_components)
return shape_model, training_shapes
def __init__(self, model, n_landmarks=0, n_align_lms=0):
super(LinearWarp, self).__init__(model)
self.pdm = OrthoPDM(model)
self.n_landmarks = n_landmarks
self.n_align_lms = n_align_lms
self.W = np.vstack((self.similarity_model.components,
self.model.components))
v = self.W[:, :self.n_dims*self.n_landmarks]
self.pinv_v = np.linalg.pinv(v)
va = self.W[:, :self.n_dims*self.n_align_lms]
self.pinv_va = np.linalg.pinv(va)
def pca_image(images, max_n_components=None):
training_shapes = []
for image in images:
training_shapes.append(image.landmarks['PTS']) # lg['all']
shape_model = OrthoPDM(training_shapes, max_n_components=max_n_components)
print(shape_model)
# visualize_pointclouds(training_shapes)
# instance = shape_model.similarity_model.instance([100., -300., 0., 0.])
# instance.view(render_axes=False)
return shape_model
def pca(path_to_images, max_n_components=None):
path_to_lfpw = Path(path_to_images)
training_shapes = []
for lg in print_progress(
mio.import_landmark_files(path_to_lfpw / '*.pts', verbose=True)):
training_shapes.append(lg) # lg['all']
shape_model = OrthoPDM(training_shapes, max_n_components=max_n_components)
print(shape_model)
# visualize_pointclouds(training_shapes)
# instance = shape_model.similarity_model.instance([100., -300., 0., 0.])
# instance.view(render_axes=False)
return shape_model
def d_dp(self, _):
r"""
The derivative with respect to the parametrisation changes evaluated at
points.
Parameters
----------
points : ``(n_points, n_dims)`` `ndarray`
The spatial points at which the derivative should be evaluated.
Returns
-------
d_dp : ``(n_points, n_parameters, n_dims)`` `ndarray`
The Jacobian with respect to the parametrisation.
"""
return OrthoPDM.d_dp(self, _)[self.n_landmarks:, ...]
def d_dp(self, _):
return OrthoPDM.d_dp(self, _)[self.n_landmarks:, ...]
def set_target(self, target):
if target.n_points == self.n_landmarks:
# densify target
target = np.dot(np.dot(target.as_vector(), self.pinv_V), self.W)
target = PointCloud(np.reshape(target, (-1, self.n_dims)))
OrthoPDM.set_target(self, target)
def d_dp(self, _):
return OrthoPDM.d_dp(self, _)[self.n_landmarks:, ...]
def set_target(self, target):
if target.n_points == self.n_landmarks:
# densify target
target = np.dot(np.dot(target.as_vector(), self.pinv_V), self.W)
target = PointCloud(np.reshape(target, (-1, self.n_dims)))
OrthoPDM.set_target(self, target)
for i in range(len(labels)):
landmarks_dataset[i] = string_array_to_np_array(labels.iloc[i])
return landmarks_dataset
# input
dataset_csv_path = "/disk1/ofirbartal/Projects/Dataset/GANeratedHands_Release/dataset_csv/test_dataset.csv"
landmarks = process_csv_input(dataset_csv_path)
start = time.time()
# PDM
training_shapes = [PointCloud(l) for l in landmarks]
shape_model = OrthoPDM(training_shapes, max_n_components=None)
shape_model.n_active_components = 0.95
end = time.time()
# shape_model = import_pickle(Path('pdm_weights.pkl'))
print(shape_model)
print("PDM Training Time: {}s".format(end - start))
export_pickle(shape_model, Path('test_pdm_weights.pkl'), overwrite=True)
# shape_model.set_target(result_landmarks) # project the target
# shape_model.target # the projected target
# draw
#create canvas on which the triangles will be visualized
# canvas = np.full([400,400], 255).astype('uint8')
#convert to 3 channel RGB for fun colors!
def __init__(self, model, transform_cls, global_transform, source=None):
self.pdm = OrthoPDM(model, global_transform)
self._cached_points = None
self.transform = transform_cls(source, self.target)
