MD-Face uses an MoE backbone plus a geometry-aware loss to achieve label-free disentangled facial attribute editing that beats unsupervised baselines, matches supervised methods, and runs faster than diffusion models.
Fda-gan: Flow-based dual attention gan for human pose transfer
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MD-Face: MoE-Enhanced Label-Free Disentangled Representation for Interactive Facial Attribute Editing
MD-Face uses an MoE backbone plus a geometry-aware loss to achieve label-free disentangled facial attribute editing that beats unsupervised baselines, matches supervised methods, and runs faster than diffusion models.