Remote Recording of Emotional and Activity Data: A Methodological Study

TL;DR

This paper introduces a new methodology for extracting high-resolution emotional and activity data from publicly available at-home exercise videos, enabling large-scale real-world studies.

Contribution

It presents a robust, accessible method for gathering emotional and physical activity data from online videos, facilitating research outside controlled environments.

Findings

Method reliably extracts high-resolution data from home videos

Source code and instructions are publicly available

Applicable for large-scale emotional response studies

Abstract

The impact of physical exercise on emotional well-being is one of the most important factors that drive sustained physical activity engagement. It is also one of the least studied topics on account of the rather elaborated setups required to quantify emotional expression during exercise. The wide adoption of at-home, physical exercise solutions has compounded this problem due to the secluded nature of these activities. We propose here a new methodology that would allow for mass emotional expression and physical activity data gathering using publicly available sources such as at-home exercise videos. We have shown that the methodology is robust enough to extract high resolution, reliable data from home videos posted on popular video share sites. The source-code and instructions for practical use are published online such that researchers can access data pertinent to emotional response during exertion in real world situations with applicability for prospective and retrospective studies.