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
Currently our data structure assumes that all individuals (animals) are tracked with the same keypoints (body parts). This assumption holds for most, but not all pose estimation datasets.
We should consider this case and decide what to do about it:
figure out a technical workaround for supporting such cases
or explicitly state that they are not supported
The text was updated successfully, but these errors were encountered:
For the time being we should probably document this more clearly. It doesn't seem to be in the Input/output section or in the API reference for the relevant functions, maybe those are nice places to put this?
Agreed, document for the short term, and think about supporting this in the long-term—in the sense of loading such data into separate movement datasets (1 per skeleton type).
Currently our data structure assumes that all individuals (animals) are tracked with the same keypoints (body parts). This assumption holds for most, but not all pose estimation datasets.
We should consider this case and decide what to do about it:
The text was updated successfully, but these errors were encountered: