Open lucasjinreal opened 2 years ago
Basically the two network architectures are the same, but I used two separate classes for separate purposes. The HMR is the one used for extracting static image features with the pretrained model weights. (only using the backbone) The Regressor is used for training TCMR.
Actually the codes are little bit modified from VIBE, so you can check their codes too:)
what's the difference between Regressor and HMR?
they seems output same thing.