Disjoint Mapping Network for Cross-modal Matching of Voices and Faces
This addresses the problem of matching voices to faces for biometric applications, offering an incremental improvement over existing methods.
The paper tackles cross-modal biometric matching of voices and faces by proposing DIMNet, which learns shared representations through individual mapping to common covariates, achieving better performance than current methods with simpler design and lower data requirements.
We propose a novel framework, called Disjoint Mapping Network (DIMNet), for cross-modal biometric matching, in particular of voices and faces. Different from the existing methods, DIMNet does not explicitly learn the joint relationship between the modalities. Instead, DIMNet learns a shared representation for different modalities by mapping them individually to their common covariates. These shared representations can then be used to find the correspondences between the modalities. We show empirically that DIMNet is able to achieve better performance than other current methods, with the additional benefits of being conceptually simpler and less data-intensive.