Feature level warping improves remote physiological measurement accuracy
FLOW: Feature-Level Optimal Warping for Generalized Remote Physiological Measurement
Computer Vision and Pattern Recognition
Summary
Measuring heart rate and other vital signs using just a camera is hard because different lighting, movement, and camera types can confuse the system. The authors created a new way called FLOW that matches and adjusts features of the input video over time to keep the measurement stable and accurate across different conditions. Their method is lightweight and can be added to current camera-based health tools without extra steps. They tested it on several datasets and found it works better than existing methods, while still capturing true physiological signals.
What this means in practice
- •For mobile health app developers: Integrate FLOW to improve heart rate monitoring reliability under varied lighting and user movement conditions in smartphone apps.$Commercial implications: Enables development of more accurate, non-contact heart rate monitors for consumer health apps, improving user experience and product competitiveness.
- •For remote patient monitoring teams: Use FLOW to enhance robustness of contactless vital sign tracking in clinical or home care settings across varying devices and environments.
Authors
Bo Zhao, Junzhe Cao, Dan Guo, Dongmin Huang, Wenjin Wang, Tao Tan, Yue Sun, Zitong YU
Abstract
Remote photoplethysmography (rPPG) enables non-contact physiological measurement but remains vulnerable to domain shifts from illumination, motion, and sensors. We propose \textbf{FLOW (Feature-Level Optimal Warping)}, an \emph{optimal transport--driven} framework for domain-generalized rPPG. FLOW integrates a \textbf{Temporal Refinement Module (TRM)} to stabilize temporal dynamics and a \textbf{Prototype-based Cross-Temporal Optimal Transport (PCOT)} module to achieve domain-invariant alignment via learnable prototypes.Beyond feature alignment, FLOW employs soft cross-temporal correspondence modeling that aligns temporal features in a flexible manner, allowing the model to respect and preserve the intrinsic rhythmic patterns of physiological signals. Moreover, the lightweight design of our modules allows seamless integration into existing end-to-end rPPG architectures without additional preprocessing. Two regularization terms further enforce source consistency and identity preservation. Theoretically, we derive a generalization bound under conditional optimal transport. Extensive experiments across four rPPG benchmarks show that FLOW achieves state-of-the-art cross-domain performance with lightweight design and strong physiological fidelity.