Related Experiment Video
Updated: Dec 8, 2025
06:45
Loneliness Assuaged: Eye-Tracking an Audience Watching Barrage Videos
Published on: May 29, 2020
4.5K
Eye-Tracking Analysis for Emotion Recognition.
Paweł Tarnowski1, Marcin Kołodziej1, Andrzej Majkowski1
1Institute of Theory of Electrical Engineering, Measurement, and Information Systems, Warsaw University of Technology, Warsaw 00-662, Poland.
Computational Intelligence and Neuroscience
|September 23, 2020
Summary
This study used eye-tracking to recognize emotions evoked by dynamic movies. Researchers achieved 80% accuracy in classifying emotional states using eye movement and pupil data with a support vector machine (SVM) classifier.
Area of Science:
- Psychology
- Computer Science
- Neuroscience
Background:
- Emotion recognition is crucial for human-computer interaction.
- Eye-tracking offers objective physiological measures of emotional responses.
- Understanding how visual stimuli evoke emotions is key to developing affective computing systems.
Purpose of the Study:
- To investigate the efficacy of eye-tracking features for emotion recognition.
- To determine the relationship between eye movement patterns, pupil diameter, and emotional states.
- To classify emotions based on physiological responses to dynamic visual stimuli.
Main Methods:
- Utilized eye-tracking data from 30 participants viewing 21 dynamic movie fragments.
- Calculated 18 features including fixations, saccades, and pupil diameter.
- Employed a support vector machine (SVM) classifier with leave-one-subject-out validation.
Main Results:
- Achieved a maximum classification accuracy of 80% for emotion recognition.
- Identified specific eye-tracking features correlating with different arousal and valence levels.
- Demonstrated the influence of movie luminance and dynamics on emotional responses.
Conclusions:
- Eye-tracking is a viable method for objective emotion recognition.
- Specific eye movement and pupil dynamics can reliably indicate emotional states.
- This research contributes to the development of emotion-aware technologies.
