A Study on Eye Tracking for Mobile Devices Using Deep Learning
Le Hoang Nam, Marco Porta · 2023
While eye tracking technology has been around for several years, it has traditionally been implemented on personal computers using specific devices. Eye tracking through smartphones or tablets is much more challenging, because it involves the use of standard cameras and typically significantly fewer computational resources. In this paper, we present a study in which, based on a large dataset of face images acquired via mobile devices, we investigate the influence of some design choices (in particular, related to optimizers, color, and regularization techniques) on a deep learning convolutional architecture. We believe that the results obtained, although preliminary, can provide a useful contribution to a challenging and constantly evolving research field such as that of eye tracking for mobile appliances.