I am looking into the joints that were ignored for the Humans 3.6m skeletons when dealing with 3D coordinates instead of angles. It is mentioned in the source code (utils/data_utils.py), line 773 that you choose to ignore the constant joints and joints at the same position with other joints:
# ignore constant joints and joints at same position with other joints joint_to_ignore = np.array([0, 1, 6, 11, 16, 20, 23, 24, 28, 31])
I can understand the reason for removing most of these joints, but what I don't get is removing all hip joints (joints 0, 1, 6, and 11). The middle hip joint is repeated (0 and 11) but both nodes were removed. Could you please elaborate what was the reason behind this decision?
I have made a sketch of what I understand of the joints numbering and their location in the skeleton. I have greyed out possible joint duplicates (in close proximity to other joints). The joints that are removed in your code using 3D coordinates are (adding 1):
[1, 2, 7, 12, 17, 21, 24, 25, 29, 32].
Hello,
I am looking into the joints that were ignored for the Humans 3.6m skeletons when dealing with 3D coordinates instead of angles. It is mentioned in the source code (utils/data_utils.py), line 773 that you choose to ignore the constant joints and joints at the same position with other joints:
# ignore constant joints and joints at same position with other joints joint_to_ignore = np.array([0, 1, 6, 11, 16, 20, 23, 24, 28, 31])I can understand the reason for removing most of these joints, but what I don't get is removing all hip joints (joints 0, 1, 6, and 11). The middle hip joint is repeated (0 and 11) but both nodes were removed. Could you please elaborate what was the reason behind this decision?
I have made a sketch of what I understand of the joints numbering and their location in the skeleton. I have greyed out possible joint duplicates (in close proximity to other joints). The joints that are removed in your code using 3D coordinates are (adding 1): [1, 2, 7, 12, 17, 21, 24, 25, 29, 32].
Thank you, Kareem.