We've never been eye to eye: A Pupillometry Pipeline for the Detection of Stress and Negative Affect in Remote Working Scenarios

Abstract

In this paper, we present a processing pipeline for the analysis of stress and negative affect based on pupillometry. We were able to show that it is possible to extract meaningful pupil features from video data recorded by an infrared- (IR-) sensitive webcam and successfully trained a Support Vector Machine on the corresponding dataset. Further, we conducted a study that shows that the proposed pipeline is suitable for the assessment of stress as well as negative affect during stress eliciting situations in a digital environment.