You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello,I am currently attempting to use other datasets to train the GMN model. I would like to understand how the motion.pkl file in the motion folder is generated. Could you provide an explanation or instructions regarding its creation?
The text was updated successfully, but these errors were encountered:
For motion.pkl, it is a pre-processed version of the raw data downloaded from the official CMU motion capture database. You can refer to another codebase of mine https://github.com/hanjq17/EGHN/blob/main/motion/preprocess.py to see the details.
In general, what you need to do to create a dataset is pretty simple: You just need to put the trajectories into a list with each trajectory being an array of shape [T, N, 3] where T is the number of timesteps for that trajectory and N is the number of nodes. Then you can refer to the Dataset object to see how this form of data can be further processed into the model input.
Hello,I am currently attempting to use other datasets to train the GMN model. I would like to understand how the motion.pkl file in the motion folder is generated. Could you provide an explanation or instructions regarding its creation?
The text was updated successfully, but these errors were encountered: