svito-zar / genea_numerical_evaluations Goto Github PK

In the calculation of the standard deviation of the jerk values, mean and std are calculated on a different level of granularity
https://github.com/Svito-zar/genea_numerical_evaluations/blob/master/calc_jerk.py#L159

Hello,
In the FGD section, what do you mean "Put challenge system results on data/Cond_??."?, And I can not find it.
Cound you please tell me some ideas?
Thanks!

Hello, what should be the data format under data/Trinity? When I execute train_AE， I unable to load the training data

Hello! Thanks for the great work!

In GENEA 2022 forked “genea-workshop / genea_numerical_evaluations”, For your bvh2npy.py, it seems that for 18 joints it will report an error KeyError, as the bvh in .zip file:

val_2022_v1_000.zip

When I experimented myself bvh2npy.npy is written like this, what do you think about this?

    target_joints = ['b_spine0', 'b_spine1', 'b_spine2', 'b_spine3', 'b_l_shoulder', 'b_l_arm', 'b_l_arm_twist',
                     'b_l_forearm', 'b_l_wrist_twist', 'b_l_wrist', 'b_r_shoulder', 'b_r_arm', 'b_r_arm_twist',
                     'b_r_forearm', 'b_r_wrist_twist', 'b_r_wrist', 'b_neck0', 'b_head']
    data_pipe = Pipeline([
        ('dwnsampl', DownSampler(tgt_fps=30, keep_all=False)),
        ('root', RootNormalizer()),
        ('jtsel', JointSelector(target_joints, include_root=False)),
        ('np', Numpyfier())
    ])