Measuring Domain Shifts using Deep Learning Remote Photoplethysmography Model Similarity
This addresses domain shift issues for rPPG models in video-based heart rate inference, offering a practical solution for in-the-wild data, though it is incremental as it builds on existing model similarity concepts.
The paper tackled the problem of domain shifts in remote photoplethysmography (rPPG) models by proposing metrics based on model similarity to measure these shifts, showing high correlation with empirical performance and achieving a 13.9% improvement in model selection without ground truth.
Domain shift differences between training data for deep learning models and the deployment context can result in severe performance issues for models which fail to generalize. We study the domain shift problem under the context of remote photoplethysmography (rPPG), a technique for video-based heart rate inference. We propose metrics based on model similarity which may be used as a measure of domain shift, and we demonstrate high correlation between these metrics and empirical performance. One of the proposed metrics with viable correlations, DS-diff, does not assume access to the ground truth of the target domain, i.e. it may be applied to in-the-wild data. To that end, we investigate a model selection problem in which ground truth results for the evaluation domain is not known, demonstrating a 13.9% performance improvement over the average case baseline.