How to visualise a motion from the dataset

I simply want to visualise a motion from the dataset without having to train a model.
From the flow of your script it is not obvious how to do it, might it be possible to elaborate on how to do it?
Thanks.

Hi @MetaDev,

You can use the forward_kinematics.py script almost as is, except you have to read from the .txt ground truth files.

Instead of

human-motion-prediction/src/forward_kinematics.py

Lines 161 to 164 in c9a2774

    
           # numpy implementation 
        
           with h5py.File( 'samples.h5', 'r' ) as h5f: 
        
             expmap_gt = h5f['expmap/gt/walking_0'][:] 
        
             expmap_pred = h5f['expmap/preds/walking_0'][:]

Do something like

  # Read data from .txt file
  data_dir = "./data/h3.6m/dataset"
  test_subject_ids = [5]
  actions = ["walking"]
  one_hot = False
  test_set, _ = data_utils.load_data( data_dir, test_subject_ids, actions, one_hot )

  subject = 5
  subaction = 1
  expmap_gt = test_set[(subject, 'walking', subaction, 'even')]
  # print( expmap_gt, expmap_gt.shape )
  expmap_pred = np.zeros_like( expmap_gt )

This will visualize the ground truth for walking.

If you want to see subjects other than 5, you will have to obtain another offset (bone lengths) array in _some_variables() from the matlab code that is provided with human3.6M.

Hope that helps!

That's perfect, thank you for the explanation.

Hi, I'm revisiting the data visualization part. I have seen that not all data is used when visualizing (only 28 joints of 32). Some joints only contain 0 values but other do seem to have motion. I tried visualizing them but the result looks weird. [image: Inline image 1] I also searched on the 3.6 M dataset website but couldn't find anything on the names of the joints in the data. Do you have an idea of what the other joints are? Thanks. Best, Harald

…

On Tue, Dec 5, 2017 at 9:42 AM, Julieta ***@***.***> wrote: Closed #21 <#21>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AG5qENF7A3jsWJR715N8n41Y63q93Vvrks5s9QHpgaJpZM4Qtf1V> .

Hi! Do you have the image? It doesn't seem to have been attached.

Re: 28 joints. I believe some of them are repeated. You may want to try plotting a number with text next to the joint and you'll see what I mean.

	# numpy implementation
	with h5py.File( 'samples.h5', 'r' ) as h5f:
	expmap_gt = h5f['expmap/gt/walking_0'][:]
	expmap_pred = h5f['expmap/preds/walking_0'][:]