search

csbhr / FFHQ-UV

The official repository of our CVPR2023 paper "FFHQ-UV: Normalized Facial UV-Texture Dataset for 3D Face Reconstruction".
MIT License
465 stars 46 forks source link

pre_model #50

Open liushifu12138 opened 1 year ago

liushifu12138 commented 1 year ago

thanks your job! Can this work train a model to do UV mapping inference? If possible, how many images are needed as a training set

csbhr commented 1 year ago

Are you referring to using Deep3D to regress the latent code z of the texture decoder? This is possible.

Using Deep3D, a self-supervised training strategy, the more high-quality training data, the better. According to the experimental details in the Deep3D paper, the number of face images in the training set is ~260K.