Quality-Aware Multimodal Fusion Reveals Implicit Identity in Valence-Arousal Features

Jisu Kim, Benjamin S. Riggan

Conventional face recognition relies on static appearance cues and degrades in unconstrained settings with expression variation, occlusion, and poor lighting. We hypothesize that a…