I think requirements.txt should be updated

[Decide02]

In fact, it's April 2024, so I think the version has changed rapidly in the meantime.

First, the first problem I faced was trying to install opencv of requirements.txt in Python 3.6 environment,

The above error has occurred. This has been resolved simply by downgrading to a version that I believe was used at the time.

pip install opencv-python==4.5.2.52

Other than that, I thought there would be no problem with requirements.txt, but when I ran it like this, the following error came out.

One interesting fact is that code that wasn't ready to run the mshot virtual environment was executed outside the virtual environment! (though other errors occurred...)

If I'm not doing anything wrong, I'll share it when I find out something new. Or if I'm doing something wrong, please help. Good luck!

[Decide02]

The above is the PHALP output value of the first frame of the video I entered.

[Decide02]

Now I could run first demo code but, when I try to run the second demo code main.py, I get the error below.

If you're having trouble getting to this point, feel free to ask questions, and if you know of a solution to the above error, please share it! Thank you!

[Decide02]

I've run the Demo code now, and I've even succeeded in seeing the mesh-coated frames based on tracklet = 1. However, I can't get the result I want... I've tried to figure out the author's intention roughly, changed the code to call the frame inside the _DEMO folder because there is no _TMP path saved in PHALP, and changed it to a key value of img_name because the key value of mask_name does not exist, but strangely, the mesh id keeps changing and not tracking properly... :cry:

# line 115
pred_masks = glob.glob(os.path.join(args.phalp_output, '_DEMO', '%s/%s*.jpg' % (imgname.split('/')[-3], frame_i[-14:-10])))
pred_masks.sort()

# line 128
if index in row_ind:
      op_id = col_ind[np.where(row_ind == index)[0][0]]
      person = json_contents['people'][op_id] 
      body_keypoints = np.reshape(person['pose_keypoints_2d'], [-1, 3]) 
      valid = 1
else:
      # Handle the case where no valid mask was found
      body_keypoints = np.zeros([25, 3])
      valid = 0
if track_frame['img_name'][index] != track_frame['frame_path'][-10:]:
      body_keypoints = np.zeros([25, 3])
      valid = 0