Open markkim1115 opened 2 years ago
Hello, These days, i'm researching based on your ProHMR idea.
I have a question. As far as I understand from the paper, z=0 mapped to a mode of body pose distribution conditioned on input image.
So, variable θ from z=0 may have highest likelihood in distribution p(θ).
But from the result, we can see the best result comes from many samplings.
Why the mode still does not give best reconstruction error? How do you think about this?
May be dumb question, forgive me. Thanks!
Hello, These days, i'm researching based on your ProHMR idea.
I have a question. As far as I understand from the paper, z=0 mapped to a mode of body pose distribution conditioned on input image.
So, variable θ from z=0 may have highest likelihood in distribution p(θ).
But from the result, we can see the best result comes from many samplings.
Why the mode still does not give best reconstruction error? How do you think about this?
May be dumb question, forgive me. Thanks!