Browsing by Subject "eye tracking"

Gaze estimation, the ability to predict where a person is looking, has become an indispensable technology in healthcare research. Current tools for gaze estimation rely on specialized hardware and are typically used in well-controlled laboratory settings. Novel appearance-based methods directly estimate a person's gaze from the appearance of their eyes, making gaze estimation possible with ubiquitous, low-cost devices, such as webcams and smartphones. This dissertation presents new methods on appearance-based gaze estimation as well as applying this technology to solve challenging problems in practical healthcare applications.

One limitation of appearance-based methods is the need to collect a large amount of training data to learn the highly variant eye appearance space. To address this fundamental issue, we develop a method to synthesize novel images of the eye using data from a low-cost RGB-D camera and show that this data augmentation technique can improve gaze estimation accuracy significantly. In addition, we explore the potential of utilizing visual saliency information as a means to transparently collect weakly-labelled gaze data at scale. We show that the collected data can be used to personalize a generic gaze estimation model to achieve better performance on an individual.

In healthcare applications, the possibility of replacing specialized hardware with ubiquitous devices when performing eye-gaze analysis is a major asset that appearance-based methods brings to the table. In the first application, we assess the risk of autism in toddlers by analyzing videos of them watching a set of expert-curated stimuli on a mobile device. We show that appearance-based methods can be used to estimate their gaze position on the device screen and that differences between the autistic and typically-developing populations are significant. In the second application, we attempt to detect oculomotor abnormalities in people with cerebellar ataxia using video recorded from a mobile phone. By tracking the iris movement of participants while they watch a short video stimuli, we show that we are able to achieve high sensitivity and specificity in differentiating people with smooth pursuit oculomotor abnormalities from those without.

We designed a novel eye tracking selection technique with progressive refinement - eye-controlled sphere-casting refined by quad-menu (EyeSQUAD) selection technique. Through a user study, we evaluated the performance of this technique with comparison of two previous selection techniques - ray-casting and SQUAD under different target size and distractor density scenarios. Results show that the EyeSQUAD technique can achieve similar selection speed as ray-casting and SQUAD and is more accurate than ray-casting although less precise than SQUAD. Finally, we summarized several insights for designing interaction techniques with eye tracking.

Joint attention is critical for social learning activities such as parent-child shared book reading. However, there is a potential disassociation of attention when the adult reads texts while the child looks at pictures. I hypothesize that the lack of joint attention limits children's opportunity to learn print-related skills. The current study tests the hypothesis with interventions that enhance real-time joint attention. Eye movements of parents and children were simultaneously tracked when they read books together on computer screens. I also provided real-time feedback to the parent regarding where the child was looking, and vice versa. Changes of dyads' reading behaviors before and after the joint attention intervention were measured from both eye movements and video records. Baseline data showed little joint attention in parent-child shared book reading. The real-time attention feedback significantly increased the joint attention and children's print-related learning. These findings supported my hypothesis that engaging in effective joint attention is critical for children to acquire knowledge and skills during shared reading and other collaborative learning activities.