Plot a 3D skeleton pose data in 3D space

vision
1

hello. Maybe it is the right place for this topic but I am working on this for a while but cannot succeed to implement it and need your help.
My goal is to reproduce this image bellow from the 3D pose data obtained from skeleton estimation of human pose using LCR-NET++
skeleton

the pose estimation for one image frame is here:
“pose3d”: [-0.013501551933586597, -0.14018067717552185, 0.03889404982328415, -0.01468866690993309, -0.052195221185684204, -0.019107796251773834, 0.1497691571712494, 0.3384685516357422, -0.01354127749800682, 0.20444869995117188, -0.01537160761654377, 0.10283246636390686, 0.16161373257637024, -0.9542085528373718, -1.0142440795898438, -0.5674616694450378, -0.6482287049293518, -0.21104587614536285, -0.26092272996902466, 0.01090222503989935, -0.06246425583958626, 0.07578188925981522, -0.06475285440683365, 0.27830997109413147, 0.16628871858119965, 0.40817680954933167, 0.4491078853607178, 0.26747873425483704, 0.3288397789001465, 0.15092524886131287, 0.14701153337955475, -0.013860990293323994, 0.31942757964134216, -0.10401999950408936, 0.2921887934207916, -0.2079567015171051, 0.12265170365571976, -0.21420519053936005, -0.07994606345891953]

Here is my code. What is missing?

from mpl_toolkits import mplot3d

import numpy as np
import matplotlib.pyplot as plt

ax = plt.axes(projection='3d')

fig = plt.figure()
xdata = np.array([-0.013501551933586597, -0.14018067717552185, 0.03889404982328415, -0.01468866690993309, -0.052195221185684204, -0.019107796251773834, 0.1497691571712494, 0.3384685516357422, -0.01354127749800682, 0.20444869995117188, -0.01537160761654377, 0.10283246636390686, 0.16161373257637024])
ydata = np.array([-0.9542085528373718, -1.0142440795898438, -0.5674616694450378, -0.6482287049293518, -0.21104587614536285, -0.26092272996902466, 0.01090222503989935, -0.06246425583958626, 0.07578188925981522, -0.06475285440683365, 0.27830997109413147, 0.16628871858119965, 0.40817680954933167])
zdata = np.array([0.4491078853607178, 0.26747873425483704, 0.3288397789001465, 0.15092524886131287, 0.14701153337955475, -0.013860990293323994, 0.31942757964134216, -0.10401999950408936, 0.2921887934207916, -0.2079567015171051, 0.12265170365571976, -0.21420519053936005, -0.07994606345891953])
ax.scatter3D(xdata, ydata, zdata, c=zdata)

plt.show()
2

I think your code is missing the logic to connect the points, as scatter3D would only visualize the points, if I’m not mistaken.
Since this question isn’t really PyTorch-related, I would guess you might get a faster answer in other discussion boards / mailing lists.
I would also check repositories working on pose detection models, as they might already have code to visualize it.