tneumann / splocs Goto Github PK

hello.

I have question about this framework.

this app works greats.

But on some dataset, It obtained crashed results.

(I used face mask dataset from Robert W. Sumner Site)

I didn't know what actaully happended here. (I think it was occurred by numerical issue..)

this is my one of my results.
video

(front face in result)

(back face in result)

Any advice would be appreciated.

thank you.

ren@keegan  /home/ren/Code/Deformation/splocs/splocs-master   python sploc.py ./data/h5/volker_aligned.h5 ./data/h5/volker_splocs.h5
/usr/local/lib/python2.7/dist-packages/numpy/core/fromnumeric.py:2645: VisibleDeprecationWarning: rank is deprecated; use the ndim attribute or function instead. To find the rank of a matrix see numpy.linalg.matrix_rank.
VisibleDeprecationWarning)
sploc.py:60: RuntimeWarning: divide by zero encountered in double_scalars
pre_scale_factor = 1 / np.std(X)
sploc.py:61: RuntimeWarning: invalid value encountered in multiply
X *= pre_scale_factor
Traceback (most recent call last):
File "sploc.py", line 170, in
args.output_anim)
File "sploc.py", line 72, in main
U, s, Vt = svd(R[:,idx,:].reshape(R.shape[0], -1).T, full_matrices=False)
File "/usr/lib/python2.7/dist-packages/scipy/linalg/decomp_svd.py", line 89, in svd
a1 = asarray_chkfinite(a)
File "/usr/local/lib/python2.7/dist-packages/numpy/lib/function_base.py", line 668, in asarray_chkfinite
"array must not contain infs or NaNs")
ValueError: array must not contain infs or NaNs

when I try to visualize the SPLOCS components, I'm receiving this error:

  File "view_splocs.py", line 86, in main
    visualization = Visualization(Xmean, tris, components)
  File "view_splocs.py", line 40, in __init__
    self.pd.point_data.normals = compute_normals(self.pd) # compute normals once for rest-shape (faster)
  File "view_splocs.py", line 18, in compute_normals
    n = tvtk.PolyDataNormals(input=pd, splitting=False)
  File "tvtk_classes/poly_data_normals.py", line 58, in __init__
  File "/usr/local/lib/python2.7/dist-packages/tvtk/tvtk_base.py", line 307, in __init__
    super(TVTKBase, self).__init__(**traits)
  File "/usr/lib/python2.7/dist-packages/traits/trait_handlers.py", line 104, in _read_only
    name, class_of( object ) )
traits.trait_errors.TraitError: The 'input' trait of a PolyDataNormals instance is 'read only'.

I find it is hard to converge， do you have any good debugging method?

First thanks for making the code public!

I am trying to use the code in geodesic.py to compute the geodesic distances for a given triangular mesh, however I am getting nan outputs. Any ideas? Many thanks in advance!

Below are some debug information I printed out:

In [361]: gg = geodesic.GeodesicDistanceComputation(vv, ff)

In [362]: gg(0)
('u0', array([1., 0., 0., ..., 0., 0., 0.]))
('u', array([3120.75912386, 1470.38703496, 1378.19606602, ..., 0. ,
0. , 0. ]))
('grad_u', array([[-5.03985517e+05, 3.43585037e+05, 2.14059926e+05],
[ 7.41695996e+03, 1.76621389e+04, -3.11673185e+04],
[ 5.52125021e+00, -8.78023615e+00, -7.32132490e+00],
...,
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00],
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00],
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00]]))
('X', array([[ 0.77964253, -0.53151033, -0.33114091],
[-0.20273969, -0.48278764, 0.85194643],
[-0.43489439, 0.69159616, 0.57668155],
...,
[ nan, nan, nan],
[ nan, nan, nan],
[ nan, nan, nan]]))
('cot1', array([0.44103681, 1.68879891, 0.60181697, ..., 0.85222178, 0.85941778,
0.98085892]))
('cot2', array([2.95458538, 0.07467702, 0.83182017, ..., 0.0895237 , 0.09506078,
0.07537141]))
('cot1', array([-0.08925637, 0.49554718, 0.34834233, ..., 0.98084458,
0.96209922, 0.87677019]))
('cot2', array([0.44103681, 1.68879891, 0.60181697, ..., 0.85222178, 0.85941778,
0.98085892]))
('cot1', array([2.95458538, 0.07467702, 0.83182017, ..., 0.0895237 , 0.09506078,
0.07537141]))
('cot2', array([-0.08925637, 0.49554718, 0.34834233, ..., 0.98084458,
0.96209922, 0.87677019]))
('Phi', array([-0.73051449, -0.72611411, -0.72558426, ..., nan,
nan, nan]))
Out[362]: array([nan, nan, nan, ..., nan, nan, nan])

Hello:
Thank you for sharing the amazing work. In the paper there is a user input binary mask feature. Could you point out how can I use that feature in your code? thank you!