Related Experiment Video
Updated: May 28, 2026
07:48
Eye Tracking During A Complex Aviation Task For Insights Into Information Processing
Published on: April 4, 2025
Combining Deep Learning Mechanisms to Predict Interests from Gaze
Aoba Izumi1, Fumiko Harada1, Hiromitsu Shimakawa1
1College of Information Science and Enginnering, Ritsumeikan University, Osaka 567-8570, Japan.
Sensors (Basel, Switzerland)
|May 27, 2026
Summary
This study introduces a deep learning model to predict personal interests from gaze and pupil diameter. This technology can personalize content recommendations and advertisements for users.
Area of Science:
- Computer Science
- Human-Computer Interaction
- Cognitive Science
Background:
- Current image processing can track gaze and pupil diameter but not infer user interests.
- Personalized user experiences are increasingly important across various digital platforms.
Purpose of the Study:
- To propose and validate a deep learning model for predicting personal interests using gaze data.
- To explore the relationship between gaze patterns, pupil dilation, and individual user interests.
Main Methods:
- Developed a deep learning model incorporating mechanisms for individual and common gaze patterns.
- Utilized pupil diameter as a training label, leveraging the property that interest increases pupil size.
- Examined model parameter modifications to understand the requirements for gaze-based interest prediction.
Main Results:
- The model successfully predicts personal interests from gaze and pupil diameter.
- Identified key parameters and data requirements for accurate interest prediction.
- Demonstrated the feasibility of using gaze for inferring subjective user interests.
Conclusions:
- Gaze tracking, combined with deep learning, offers a novel method for inferring personal interests.
- This approach has significant potential for enhancing user experiences through personalized content and services.
- Further research can refine the model for broader applications in digital environments.
