Quantifying Interest from Facial Images, and the Role of Video in Effective Business Calls

Abstract

It is said, to succeed in business, put the interest of customers ahead of your own. This is great advice to businesses, racing to deploy video calls to provide COVID-safe meetings and/or benefit from general conveniences and gain productivity. If 55% of the communication is truly nonverbal as suggested by Prof. Albert Mehrabian from University of California in Los Angeles, putting the interest of the customer ahead means listening carefully to the nonverbal channels. In video calls, one must try hardest to read, understand and manage the customer interest expressed in the video images.In this study, our primary objective was to analyse and evaluate the practicality of a project to develop a machine-learning model that can predict and quantify a level of interest of a video caller from his facial images. As such, our secondary objective was to explore facial and behavioural expressions, highly correlated to or triggered by interests that are confirmed by imaged human subjects themselves, as well as explore areas and methods to capture them as to generate machine-learning training data.The result suggested the project to develop a machine-learning model would be practical. The finding included that a model based on static facial features visible in a video frame could be possible, but a model based on moving facial features estimated from sequence of consecutive video frames could do better, especially those with acute focus on eye movements.

Keywords: video call communication conference zoom quantify interest digital transformation sales telecommute face facial image eye gaze blink emotion sentiment

DOI: 10.54941/ahfe1003238