Predicting Team Well-Being through Face Video Analysis with AI

Abstract

Well-being is one of the pillars of positive psychology, which is known to have positive effects not only on the personal and professional lives of individuals but also on teams and organizations. Understanding and promoting individual well-being is essential for staff health and long-term success, but current tools for assessing subjective well-being rely on time-consuming surveys and questionnaires, which limit the possibility of providing the real-time feedback needed to raise awareness and change individual behavior. This paper proposes a framework for understanding the process of non-verbal communication in teamwork, using video data to identify significant predictors of individual well-being in teamwork. It relies on video acquisition technologies and state-of-the-art artificial intelligence tools to extract individual, relative, and environmental characteristics from panoramic video. Statistical analysis is applied to each time series, leading to the generation of a dataset of 125 features, which are then linked to PERMA (Positive Emotion, Engagement, Relationships, Meaning, and Accomplishments) surveys developed in the context of positive psychology. Each pillar of the PERMA model is evaluated as a regression or classification problem using machine learning algorithms. Our approach was applied to a case study, where 80 students collaborated in 20 teams for a week on a team task in a face-to-face setting. This enabled us to formulate several hypotheses identifying factors influencing individual well-being in teamwork. These promising results point to interesting avenues for research, for instance fusing different media for the analysis of individual well-being in teamwork.

No pain no gain - Giving real-time emotional feedback in a virtual mirror improves collaboration in virtual teamwork

“…The Facial Analysis System (FAS) [62] analyzed video recordings to detect seven discrete emotions (neutral, surprised, happy, fearful, disgusted, angry, sad), along with 3D head poses and brightness levels. This system provided VAD-values (valence, arousal, dominance) and head motion patterns, contributing to a comprehensive understanding of non-verbal cues and interpersonal dynamics within the team, see Table 1.…”

“…To ensure the robustness of the analysis, both datasets, which were susceptible to outliers due to their nature of collection, were initially prepared for further processing [62,63]. Gaussiandistributed features were standardized using scikit-learn to ensure they have a mean of zero and a standard deviation of one [64].…”