Quantifying Dwell Time With Location-based Augmented Reality: Dynamic AOI Analysis on Mobile Eye Tracking Data With...

Area of Science:

• Computer Vision
• Human-Computer Interaction
• Cognitive Science

Background:

• Mobile eye tracking offers valuable egocentric vision data for naturalistic studies but suffers from noise and analysis challenges with moving targets.
• Current analysis tools are limited, often necessitating time-consuming manual annotation, hindering the scalability of mobile eye-tracking research.
• Nonlinear movement and object disappearances in outdoor settings complicate automated area of interest analysis.

Purpose of the Study:

• To introduce an automated method for analyzing mobile eye-tracking data, specifically addressing challenges with moving targets in naturalistic environments.
• To improve the efficiency and accuracy of processing noisy eye-tracking data from outdoor studies.
• To enable more scalable and less labor-intensive analysis of mobile eye-tracking data.

Main Methods:

• A fine-tuned Vision Transformer (ViT) model was developed for classifying frames containing gaze markers.
• The ViT model was trained on a manually labeled subset (1.98%) of the entire dataset over three epochs.
• The model's performance was evaluated using hold-out data to assess its accuracy.

Main Results:

• The fine-tuned Vision Transformer model achieved a high accuracy of 99.34% on hold-out data.
• The method was successfully applied to quantify participant dwell time on a tablet during an outdoor augmented reality biodiversity education application test.
• Demonstrated the model's capability to handle complex, naturalistic scenarios with moving elements.

Conclusions:

• The developed Vision Transformer-based method offers an accurate and efficient solution for analyzing mobile eye-tracking data in naturalistic studies.
• This approach significantly reduces reliance on manual annotation, paving the way for broader application of mobile eye tracking.
• The method shows potential for application in diverse research areas requiring analysis of gaze behavior in dynamic environments.