search

themattinthehatt / behavenet

Toolbox for analyzing behavioral videos and neural activity
https://behavenet.readthedocs.io/
MIT License
57 stars 15 forks source link

What is the shape of the 'labels' dataset in the HDF5 file? #13

Closed cxrodgers closed 4 years ago

cxrodgers commented 4 years ago

I'm following these instructions to make my HDF5 file: https://behavenet.readthedocs.io/en/latest/source/data_structure.html#data-structure

I know that there is also a "labels" dataset with joint coordinates. What is the shape of this suppose to be? Is it (n_frames, 2, n_labels), with the second axis being "x" and "y" coordinates? Or (n_frames, n_labels, 2)?

thanks!

cxrodgers commented 4 years ago

By trial and error, it seems like it is just (n_frames, 2 * n_joints)!

themattinthehatt commented 4 years ago

Hey Chris, if you look in the bottom part of the left column on readthedocs you should see a button that says v:latest; if you click on this you can change which code version you are referencing. Switch over to the develop branch and you'll get updated documentation, which includes some info on the labels.

That being said, I just realized that even the label documentation isn't super clear, so I'll make sure to update that!

themattinthehatt commented 4 years ago

@cxrodgers just made a small update to the data structure documentation in the develop branch - nothing new for you there, but if you run into any other issues that aren't covered in the documentation please let me know!