ZhengZerong/THUman4.0-Dataset

Evaluation scheme

Opened this issue · 0 comments

initialneil commented

Thanks for the dataset. I'm writing to compare with your paper.
May I ask about the evaluation scheme?
The comparison is against Table 1 of your ECCV 2022 paper. You mentioned that training is on some frames and testing is on others.

  1. Hoody is subject00 right? Jacket is which subject?
  2. What is the list of training frames and testing frames?
  3. Is the PSNR calculated on the masked region only? Or is it calculated on the whole image (with black background)?