Closed fanchunpeng closed 2 years ago
Thanks for asking this question. The reported results of both v1 (cvpr workshop) and v2 (iccv) papers are from the current version of implementation. We made a mistake when we wrote v1 version: the facial component was actually not added to the body representation due to a typo in the messy code.
Thus, in the v2 version, we corrected this and clarified some frequently asked implementation details. There is only one version of implementation (which is this repo), although we have two versions of paper. Sorry for this confusion. Just a side note, in our sanity check, with or without the facial component in the body representation, there is no big difference in the result.
Please get M(fg) as the union of the masks of all pose-transferred body segments as implemented in this repo.
Hello! The V1 and V2 versions of the paper differ greatly in Body representation . What kind of implementation effect is good? How do I get M (fg) in V1? Are the current implementations implemented according to the latest V2? Thank you for your work.