Question about adapting V-JEPA to 3D imaging

[arnaudberenbaum]

Hello everyone, Thanks for the source code ! Is it possible to adapt V-JEPA to 3D imaging (like medical imaging)? I don't know which aproach would work best between converting imaging to pseudo-short movies of 2d slices or to adapt Vision transformer model to accept volumes instead of 2d imaging. Thanks a lot :)