Presentation
A Differential 2D Gaussian Ellipse Based Eye Movement Analysis
DescriptionAs data-driven analysis methods powered by artificial intelligence have matured, research on visual attention prediction has advanced markedly. However, gaze-point data directly acquired from eye-tracking devices are often characterized by high noise levels. Consequently, it is essential to develop an eye-movement data model that more accurately reflects real-world driver visual behavior. In this paper, we propose the Differential 2D Gaussian Ellipse (D2DGE) representation, which considers the distribution of gazes within a specific time window and thus eliminates the noise introduced by both the eye-tracking devices and the unconscious random gazes during driving. To validate D2DGE, a Generative Adversarial Imitation Learning model was trained on both raw gaze data and the D2DGE data. The proximity of the generated data to the original data was compared based on the Kullback-Leibler divergence metric. Then, we compared the five key metrics of D2DGE representation between novice and experienced drivers. The results showed that the D2DGE data can better approximate the raw data, and the D2DGE data contained richer information compared to raw gaze data. The findings indicate that the D2DGE can be a promising alternative to describe gaze distribution during driving.
Event Type
Industry/Practitioner Content
Lecture
TimeFriday, October 17th8:40am - 9am CDT
LocationGrand A
Perception and Performance
Similar Presentations