Blending camera and 77 GHz radar sensing for equitable, robust plethysmography

With the resurgence of non-contact vital sign sensing due to COVID-19 pandemic, remote heart-rate monitoring has gained significant prominence. Many existing methods use cameras; however previous work shows a performance loss for darker skin tones. In this paper, we show through light transport analysis that camera modality is fundamentally biased against We propose reduce bias multi-modal fusion with complementary and fairer - radar. Through novel debiasing oriented framework, achieve gains over all tested baselines tone fairness improvements RGB modality. That is, associated Pareto frontier between improved when compared addition, are obtained radar-based method, small trade-offs in fairness. also open-source largest estimation dataset paired radar measurements focus on representation.