A method called STCCL uses learned spatial and temporal correlation metrics between facial regions to supervise expression changes while preserving speech animations without paired training data.
Contrastive decoupled representation learning and regularization for speech-preserving facial expression manipulation,
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
-
Learning Spatial-Temporal Coherent Correlations for Speech-Preserving Facial Expression Manipulation
A method called STCCL uses learned spatial and temporal correlation metrics between facial regions to supervise expression changes while preserving speech animations without paired training data.