Issues when running sandbox version of 3d pose baseline with posenet-python ? (setting an array element with a sequence. /home/tm/pf/posenet-python-final/000000000000_keypoints.json setting an array element with a sequence.)

Hi , after porting the code of posenet-python https://github.com/rwightman/posenet-python to fit according to the 3d-pose.baseline. I am unable to get predicted 3d pose . The sample output of the 2d joint detections are as follows . (for one of the files)

openpose_3dpose_sandbox as input

{"people": [{"pose_keypoints_2d": [[65.83929879324776, 225.61339058925535, 59.41166279478843, 228.55210823006283, 60.82692164071598, 222.30894846404712, 60.806538362680755, 231.52718520576977, 62.76157592394337, 212.95569881188416, 82.17437222877645, 236.26046220220908, 80.74839289884389, 215.0094041362353, 99.23341810333062, 248.18655674680292, 94.74878248961075, 203.9242716924542, 97.9518383334142, 241.0167779707991, 90.32739378650736, 203.03192621778862, 120.93708346349112, 234.44128364633934, 119.30266528396132, 216.48404405356277, 155.9352742070737, 244.5989955321315, 151.30534438613037, 206.35645879263694, 194.62511536497507, 247.81470595099108, 188.79508184350055, 203.35007549404685, 0.0, 0.0]]}]}

The error message is as follows for realtime visualzation

  File "src/openpose_3dpose_sandbox_realtime.py", line 237, in <module>
    tf.app.run()
  File "/home/tm/.local/lib/python3.5/site-packages/tensorflow/python/platform/app.py", line 125, in run
    _sys.exit(main(argv))
  File "src/openpose_3dpose_sandbox_realtime.py", line 83, in main
    logger.debug("found {0} for frame {1}".format(xy, str(frame)))
KeyboardInterrupt

And the following for openpose_3dpose_sandbox.py the following is noted

  File "src/openpose_3dpose_sandbox.py", line 232, in main
    smoothed = read_openpose_json()
  File "src/openpose_3dpose_sandbox.py", line 129, in read_openpose_json
    drop_curves_plot = show_anim_curves(cache, plt)
  File "src/openpose_3dpose_sandbox.py", line 32, in show_anim_curves
    y = val[:,o+1]
IndexError: index 1 is out of bounds for axis 1 with size 1

[basicvisual]

@ArashHosseini , i think i might have noticed another problem in my implementation. I have not arranged the 2D keypoints , to human 3.6 order. The keypoint order for posenet is as follows

https://github.com/tensorflow/tfjs-models/blob/master/posenet/src/keypoints.ts

For Posenet 'nose', 0 'leftEye', 1 'rightEye', 2 'leftEar', 3 'rightEar', 4 'leftShoulder', 5 'rightShoulder', 6 'leftElbow', 7 'rightElbow', 8 'leftWrist', 9 'rightWrist', 10 'leftHip', 11 'rightHip', 12 'leftKnee', ' 13 rightKnee', 14 'leftAnkle', 15 'rightAnkle' 16

Currently looking into that .

[basicvisual]

@ArashHosseini , do you know the reason why the image is on another plane ? To comply with open pose i just took the mean , of right shoulder (x1,y1) and left shoulder (x2,y2) and rearranged as per the open pose keypoint ordering

https://github.com/una-dinosauria/3d-pose-baseline/issues/128#issue-444309532