Automatic analysis of in-the-wild mobile eye-tracking experiments
Stijn De Beugher, Geert Brône, Toon Goedemé · Lirias (KU Leuven) · 2016
We discuss a novel method for the analysis of mobile eye- tracking data in natural environments. Mobile eye-trackers generate large amounts of data, making manual analysis very time-consuming. Available solutions, such as marker-based analysis minimize the manual labour, however they require experimental control, making real-life experiments practically infeasible. Here, we discuss a novel method for the processing of mobile eye-tracking data based on the integration of computer vision techniques. Using such an approach allows us to automatically detect specific objects, faces and human bodies in images captured by a mobile eye-tracker. By mapping the gaze data on top of those detections, we get insights in the visual behaviour. In this paper, we briefly describe the applied image processing techniques. We also present a thorough comparison between the manual analysis and our automatic analysis in both speed and accuracy on a challenging, large-scale real-life experiment.