Sign In
insight - Heterogeneous Face Recognition
Bridging the Domain Gap in Heterogeneous Face Recognition through Conditional Adaptive Instance Modulation
The core message of this paper is that the domain gap in Heterogeneous Face Recognition (HFR) can be effectively addressed by conceptualizing different modalities as distinct styles and employing a Conditional Adaptive Instance Modulation (CAIM) module to seamlessly adapt the intermediate feature maps of a pre-trained face recognition network.
Enhancing Pretrained Face Recognition Models for Heterogeneous Face Matching Using Domain-Invariant Units
A novel approach for learning domain-invariant layers called Domain-Invariant Units (DIU) to enhance pretrained face recognition models for the task of heterogeneous face recognition.