Heterogeneous Face

Heterogeneous face recognition (HFR) focuses on accurately matching facial images acquired from different modalities (e.g., visible light, near-infrared, thermal), overcoming the significant domain discrepancies between them. Current research emphasizes developing modality-agnostic models, often leveraging pre-trained face recognition networks and incorporating modules like conditional adaptive instance modulation or domain-invariant units to bridge the domain gap, even with limited paired training data. This field is crucial for enhancing the robustness and applicability of face recognition systems in diverse and challenging real-world scenarios, such as security and surveillance.