Hello,
I’m interested in the TTM task in the Social Interactions, and
I do want to join the challenge. However, I have some problems
with the training subset.
Since Social Interactions and AV Diarization share the same
dataset, inputs of tasks inSocial Interactions (bounding box)
become the targets of tasks in AVD. It seems that in the
previous year this was solved by giving the processed face
crops and wav files (ref)
Will there be any changes this year? I thought the current
setting is quite limited and it is a bit weird to have only the
target faces when it comes to pratical scenarios.