KCS matrix seems to be incorrect

[josecanedo]

Excellent paper and code! Thanks.

By analyzing the KCS matrix and reassembling the skeleton following the paper i got something like:
6 -> 7 -> 8 -> 9
7-> 10 -> 11 -> 12
7-> 13 -> 14 -> 15
which seems to be the upper body starting at the root (hip joint), but the lower bones assemble like this:
0 -> 6
0 -> 1 -> 2
3 -> 4 -> 5
I was expecting something like this:
6 -> 0 -> 1 -> 2
6 -> 3 -> 4 -> 5

Is there a error in the KCS matrix or I misinterpreted the skeleton?

KCS matrix

Ct = tf.constant([
      [1., 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1],
      [-1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
      [0, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
      [0, 0, 1, 0, 0, 0, 0, 0, 0, 0 , 0, 0, 0, 1, 0],
      [0, 0, -1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
      [0, 0, 0, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
      [0, 0, 0, 0, 1, 0, 0, 0, 0, 0 , 0, 0, 0, 0,-1],
      [0, 0, 0, 0, -1, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0],
      [0, 0, 0, 0, 0, -1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
      [0, 0, 0, 0, 0, 0, -1, 0, 0, 0, 0, 0, 0, 0, 0],
      [0, 0, 0, 0, 0, 0, 0, -1, 1, 0, 0, 0, 0, 0, 0],
      [0, 0, 0, 0, 0, 0, 0, 0, -1, 1, 0, 0, 0, 0, 0],
      [0, 0, 0, 0, 0, 0, 0, 0, 0, -1, 0, 0, 0, 0, 0],
      [0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ,-1, 1, 0, 0, 0],
      [0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  0,-1, 1, 0, 0],
      [0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  0, 0,-1, 0, 0]])

[Aratrik]

Strangely the last but one column has two ones, rather it should have a 1 and -1. I matched the matrix with standard MPII names I found it to be correct otherwise.

[lvandoit]

I noticed that in the discriminator these was a wassertein loss. But in the code it is defined as 'mean(y_true * y_pred)'. So there had any precess for it. @Aratrik @josecanedo

[CHIKKIZZY]

Strangely the last but one column has two ones, rather it should have a 1 and -1. I matched the matrix with standard MPII names I found it to be correct otherwise.

I count exactly one pair of 1 and -1 in each column. You may have counted along the row?