Contactless Heart Rate and HRV Estimation from Neck Videos

This repository contains the implementation code for the paper:

Contactless Heart Rate and Heart Rate Variability Estimation from Neck Videos
Mohammad Muntasir Rahman, Amirtahà Taebi
Presented at the 2025 IEEE EMBC which was hold in Copenhagen, Denmark
[Paper Link] (link will be updated once available)

Overview

This project introduces a non-contact, video-based approach for estimating heart rate (HR) and heart rate variability (HRV) from short videos of the human neck. We evaluated six widely-used video-based pulse extraction methods:

• GREEN
• CHROM (Chrominance-based)
• POS (Plane-Orthogonal-to-Skin)
• OMIT (Orthogonal Matrix Image Transformation)
• ICA (Independent Component Analysis)
• LGI (Local Group Invariance)

All extracted signals were validated against synchronized ECG recordings. We also developed a robust, adaptive peak detection algorithm for estimating HR and HRV in noisy video-based signals.

Contents

• Run rPPG_signal_extraction.py to extract the video-based pulse signal from the neck region.
• Use rPPG_signal_visualization.m to preprocess the signal, HR and HRV, and visualize the results.

Requirements

• MATLAB (R2022a or later)
• Python

Dependencies adapted from:

• 📦 iPhys Toolbox
• 📦 rPPG-Toolbox

Acknowledgments

This work was supported by:

• National Science Foundation (NSF) — Grant No. 2340020
• SMART Business Act Grant — Grant No. 2024-04
  (Mississippi Institutions of Higher Learning)

Citation

If you use this code in your work, please cite:

@inproceedings{rahman2025contactless,
  title={Contactless Heart Rate and Heart Rate Variability Estimation from Neck Videos},
  author={Rahman, Mohammad Muntasir and Taebi, Amirtahà},
  booktitle={Proceedings of the IEEE EMBC 2025},
  year={2025}
}