Browser Gaze and Emotion Detection

Abstract

In recent years, more and more companies have switched their activities to the online area. As a result, the website is the central place where the company interacts with customers. By analyzing page views, companies can determine which pages are visited frequently or how long a visitor lingers on the page. The goal of this work is to go one step further and provide companies with a tool that allows them to gain a better understanding of their customers. We analyze existing methods that track the gaze of the user visiting a web page by using a standard webcam and build a gaze estimator that runs entirely in the browser. No server back-end is used, this way no sensitive user data leaves the user‘s device and we can protect the privacy of the user. We specifically focus on building a new gaze estimator that achieves high accuracy in real-world use cases and not just in controlled environments. We further extend the capabilities of our tool by integrating simple facial emotion recognition. The collected data can then be forwarded anonymously to the company by the user. The data can be analyzed and can be used to understand which elements of the web page catch the eye of users and leave a positive or negative impression. Together with the information that is currently being collected from the web pages, it is possible for companies to gain a better understanding of their customers and how they interact with their website.