StreamPPG: Low-Latency rPPG Estimation via Consistent Privileged Learning
2026-06-22 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed StreamPPG, a method to measure heart signals from facial videos quickly and without contact. Unlike previous methods that either take a long time or miss important heartbeat patterns, their approach works frame-by-frame while still being accurate. They trained their model using a special technique that uses real heartbeat data to help it learn better. Tests showed StreamPPG is both fast enough for real-time use and accurate across different datasets.
remote photoplethysmographyblood volume pulsephysiological signal estimationframe-wise approachclip-wise approachconsistent privileged learningreal-time monitoringfacial videosheartbeat detectionedge devices
Authors
Yiming Li, Yihan Yang, Yuguang Chu, Yuanhui Hu, Si-Yuan Cao, Xiaohan Zhang, Xiaokai Bai, Zhe Wu, Hui-Liang Shen
Abstract
Remote photoplethysmography (rPPG) estimates the blood volume pulse (BVP) signal from facial videos, enabling contact-free health monitoring. Conventional clip-wise approaches, which use video clips as input, require capturing over one hundred frames before inference, thus introducing several seconds of delay and hindering real-time use. Meanwhile, frame-wise approaches struggle to capture long-range temporal and periodic features of physiological rhythms, and therefore lead to reduced estimation accuracy. To overcome these issues, we propose StreamPPG, a unified architecture that enables low-latency frame-wise physiological signal estimation while achieving competitive accuracy compared with clip-wise approaches. StreamPPG is trained under a consistent privileged learning (CPL) strategy, which leverages ground-truth rPPG signals as privileged information to enhance the model's representation capability. Extensive experiments demonstrate that StreamPPG achieves state-of-the-art accuracy across multiple datasets while maintaining real-time throughput on edge devices.