Architectural Tricks for Deep Learning in Remote Photoplethysmography

TL;DR

This paper explores architectural enhancements for convolutional neural networks to improve heart rate estimation from facial color signals in stationary and motion scenarios, using classification and combined loss functions.

Contribution

It introduces specific architectural tricks and a combined loss function to enhance deep learning performance in remote photoplethysmography.

Findings

Improved HR estimation accuracy in stationary scenarios.

Enhanced robustness of HR estimation during motion.

Convolutional filtering layers and combined loss function boost performance.

Abstract

Architectural improvements are studied for convolutional network performing estimation of heart rate (HR) values on color signal patches. Color signals are time series of color components averaged over facial regions recorded by webcams in two scenarios: Stationary (without motion of a person) and Mixed Motion (different motion patterns of a person). HR estimation problem is addressed as a classification task, where classes correspond to different heart rate values within the admissible range of [40; 125] bpm. Both adding convolutional filtering layers after fully connected layers and involving combined loss function where first component is a cross entropy and second is a squared error between the network output and smoothed one-hot vector, lead to better performance of HR estimation model in Stationary and Mixed Motion scenarios.