Presentation
Evaluating the performance of eye movement detection algorithms across tasks that elicit different eye movement types in virtual reality.
SessionPoster Session 2
DescriptionVirtual Reality (VR) technology offers promising tools for Human Factors and Ergonomics research, such as eye tracking capabilities that enable researchers to collect participants’ eye movements while completing a task. To analyze eye movement data, researchers commonly apply eye detection algorithms to identify eye movements, such as eye fixations, saccades, and smooth pursuits, that occurred throughout the task. However, a task might elicit all or only some of these eye movements from participants, which can make selecting an appropriate algorithm a challenge, as many algorithms are designed to only identify specific eye movement types. As a result, the present exploratory study evaluated the performance of two algorithms (I-VVT and I-VDT), designed to identify eye fixations, saccades, and smooth pursuits, across two tasks that elicited smaller subsets of eye movement types in VR. Specifically, one task elicited eye fixations and saccades, while the other only elicited smooth pursuits. Our results showed that the I-VDT algorithm outperformed the I-VVT algorithm. However, neither algorithm accurately identified all the eye movements that occurred, as performance had to be balanced across tasks. Thus, future research should investigate whether multiple algorithms could be applied to identify different eye movements within a task to increase performance.
Event Type
Poster
TimeWednesday, October 15th5:30pm - 6:30pm CDT
LocationRiverside East
Similar Presentations