A GAN-based method generates facial attributes directly on 3D face UV maps from posed images using texture completion and attribute synthesis networks, claiming better accuracy and identity preservation than prior 2D approaches.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Pose-variant 3D Facial Attribute Generation
A GAN-based method generates facial attributes directly on 3D face UV maps from posed images using texture completion and attribute synthesis networks, claiming better accuracy and identity preservation than prior 2D approaches.