Object recognition and person detection for mobile eye-tracking research: A case study with real-life customer journeys

The recent development of user-friendly plug-and-play mobile eye-tracking technology has paved the way for research into visual behavior and real-life user experience in natural environments, such as public spaces, commercial environments or interpersonal communicative settings. The challenge for this new type of pervasive eye-tracking is the processing of data generated by the systems used in real-world environments [7]. Recently, several solutions to the analysis challenge have been proposed (see [2] for an overview). The best-known technique is the use of markers (infrared or natural) to predefine potential areas of interest (AOI), generating a two-dimensional plane within which eye gaze data can be collected for longer stretches of time and generalized across subjects. This paper presents an alternative to the AOI-based methods, building on recent studies combining object recognition algorithms with eye-tracking data [1] and [7].