Motion Matters: Neural Motion Transfer for Better Camera Physiological Measurement

WACV 2024

Oral [Top 2.6%]

1UNC Chapel Hill, 2University of Washington
*denotes equal advising

Examples of motion augmentation as applied to the UBFC-rPPG dataset. The driving videos utilized are from the TalkingHead-1KH dataset.

Our neural motion augmentation pipeline for the task of remote PPG estimation reduces error in heart rate estimation by up to 79% in inter-dataset results using TS-CAN and 47% over existing results using SOTA methods on PURE.

Abstract

Machine learning models for camera-based physiological measurement can have weak generalization due to a lack of representative training data. Body motion is one of the most significant sources of noise when attempting to recover the subtle cardiac pulse from a video. We explore motion transfer as a form of data augmentation to introduce motion variation while preserving physiological changes of interest. We adapt a neural video synthesis approach to augment videos for the task of remote photoplethysmography (rPPG) and study the effects of motion augmentation with respect to 1. the magnitude and 2. the type of motion.

After training on motion-augmented versions of publicly available datasets, we demonstrate a 47% improvement over existing inter-dataset results using various state-of-the-art methods on the PURE dataset. We also present inter-dataset results on five benchmark datasets to show improvements of up to 79% using TS-CAN, a neural rPPG estimation method. Our findings illustrate the usefulness of motion transfer as a data augmentation technique for improving the generalization of models for camera-based physiological sensing. We release our code for using motion transfer as a data augmentation technique on three publicly available datasets, UBFC-rPPG, PURE, and SCAMPS, and models pre-trained on motion-augmented data on our project page.

Pipeline

Key Results

Please refer to our pre-print for the full results, including ablation studies.

Additional Materials

In addition to the motion augmentation pipeline code, we provide pre-trained models, motion analysis scripts that leverage OpenFace, and a Google Drive link to the driving videos that we used for our experiments in our GitHub repository.

BibTeX

@misc{paruchuri2023motion,
      title={Motion Matters: Neural Motion Transfer for Better Camera Physiological Sensing}, 
      author={Akshay Paruchuri and Xin Liu and Yulu Pan and Shwetak Patel and Daniel McDuff and Soumyadip Sengupta},
      year={2023},
      eprint={2303.12059},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}