Eye tracking and game mechanics: An evaluation of unmodified tablet computers for use in vision screening software for children

Abstract

Binocular vision issues are a primary cause of reading and writing difficulties among children, with as many at 25 percent of school-aged children needing vision therapy. Vision issues are a better predictor of academic success and quality of life than race or socioeconomic status. In spite of validated screening exams and vision therapy to address these issues, the vast majority of children are never examined, or if examined do not receive follow up care. Barriers to access include financial limitations, availability of qualified eye care professionals and child engagement in screening and therapy. Current screening and therapy methodologies are not scalable, but computer programs and mobile applications already deployed in medical and behavioral fields offer insight to the impact that new screening and therapy technologies could provide. However, potential negative effects from the prolonged used of tablet screens must also be considered. A study was conducted to evaluate the use of computer vision and machine learning algorithms to estimate user gaze when applied to video input from an unmodified tablet computer. Accurate gaze estimation would enable the development of tablet-based vision screening and therapy applications. Recommendations for future work are made based on the results of this study.