Zongwei97/UnTrack

About mixed RGB-D/T/E data training?

laisimiao opened this issue · 1 comments

May I know during UnTrack training, it does need three modalities data simultaneously or any combination of modalities can be used as input?

Hi,

To train the model, we mix the D/T/E modality in the minibatch.

In such a way, we enable cross-modal interaction.