Ik geloof mijn ogen niet: eye tracking voor kmo's en non profit

What are people looking at? Scientists use eye-trackers to answer this question. These devices register eye movements to determine how fast, how often, and how long people are looking at something. While eye-tracking used to be bound to lab environments, the recent development of eye-tracking glasses (also known as mobile eye-tracking devices) enables observing viewing behavior ‘into the wild’. The main goal of this project was to make mobile eye-tracking more accessible for SMEs and non-profit organizations. In particular, we (1) investigated best practices when conducting experiments involving mobile eye-tracking and (2) explored how the main constraint of using mobile eye-tracking (lengthy and hence expensive data analysis) can be averted.

Concerning the first goal (1), we conducted several experiments involving mobile eye-tracking in different settings, such as a supermarket, a waste management facility, and a police school. In doing so, we came up with several best practices concerning both the set-up (recruitment, potential issues, and guidelines concerning incidence of light and viewing distance), and analysis of eye-tracking data (how long does the analysis take and which parameters are most relevant).

The second goal (2) concerned finding a solution for the high cost of analyzing mobile eye-tracking data, which is currently the main factor that impedes most businesses from using it. The reason behind this elevated cost is the fact that mobile eye-tracking data is analyzed by manual coding. In this project, we worked together with EAVISE (KU Leuven) who developed a software that was designed to automate the coding process. In particular, the software uses computer vision algorithms to recognize and automatically label objects in eye-tracking recordings. In spite of the great progress made, the software is not viable yet. One of the main stumbling blocks is the low resolution of the eye-tracking recordings, which impedes correct recognition and labelling.