Eye-tracking art

Abstract

Eye tracking provides an interface in which a camera records the user’s eye and obtains their gaze point with the use of image processing techniques. However, eye tracking technology may face certain challenges which makes it difficult to provide the user’s intended interpretation of line art accurately. In this thesis, we have used Pupil Labs, an open-source publicly available eye tracking device in order to depict line art sculptures. In order to improve the output originally presented by Pupil, we have utilised the front-facing camera of Pupil to obtain an image of the sculpture the user is perceiving. We then retrieved points of interest of this image, and matched them with the user’s gaze points, and removed any rapid fixation changes which have been caused unintentionally by the user. The results provided indicate that from the techniques which have been tested, there has been a positive improvement on the output which is originally provided by Pupil, with the aims and objectives set being accomplished. This problem can be further improved in the future by experimenting with different techniques, possibility of automating testing with collection of sufficient data, and enhancing dimensionality to the data utilised by adding accessories to the Pupil headset.