DisFace3DNet: Explainable Facial Attractiveness Prediction via 3D Component Disentanglement
We propose DisFace3DNet, which uses 3D component disentanglement to learn seven component reference scores from overall ratings with auxiliary weak semantic supervision, without human-labeled component targets.
ProofPaper ↗
Key points
- Facial attractiveness prediction usually assigns one overall rating, leaving the roles of shape, appearance, and viewing conditions implicit.
- Designated 3D representations and image cues feed jointly learned routes for identity, skin, hair, light, background, expression, and pose.
- Skin, hair, and facial shape account for the largest component-wise prediction variation.
- DisFace3DNet thus connects overall prediction to quantitative analysis of the facial and contextual cues entering each estimate.
Sources (1)
- [1]DisFace3DNet: Explainable Facial Attractiveness Prediction via 3D Component DisentanglementarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 10:33 AM
We propose DisFace3DNet, which uses 3D component disentanglement to learn seven component reference scores from overall ratings with auxiliary weak semantic supervision, without human-labeled component targets.
Facial attractiveness prediction usually assigns one overall rating, leaving the roles of shape, appearance, and viewing conditions implicit.
Extractive summary: sentences quoted from the sources.