Related Experiment Video
Updated: Oct 12, 2025
08:25
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
9.2K
Mobile Eye-Tracking Data Analysis Using Object Detection via YOLO v4
Niharika Kumari1, Verena Ruf1, Sergey Mukhametov1
1Physics Education Research Group, Physics Department, TU Kaiserslautern, 67663 Kaiserslautern, Germany.
Sensors (Basel, Switzerland)
|November 27, 2021
Summary
This study introduces object recognition for mobile eye tracking in labs. YOLOv4 with optical flow offers fast, accurate object detection, simplifying data analysis and enabling real-time responses.
Area of Science:
- Educational Technology
- Computer Vision
- Human-Computer Interaction
Background:
- Remote eye tracking is crucial for analyzing learning processes in real-world settings.
- Mobile eye trackers offer greater flexibility than stationary ones but face analysis challenges.
- Manual analysis of mobile eye-tracking data is time-consuming and impractical for dynamic environments.
Purpose of the Study:
- To explore using object recognition models for assigning mobile eye-tracking data to real objects.
- To evaluate the efficiency and accuracy of different Convolutional Neural Networks (CNNs) for this task.
- To simplify the analysis of mobile eye-tracking data in authentic student lab courses.
Main Methods:
- Comparison of three CNN models: Faster Region-Based CNN, YOLOv3, and YOLOv4.
- Integration with optical flow estimation for enhanced object detection.
- Application within an authentic student laboratory course setting.
Main Results:
- YOLOv4, combined with optical flow estimation, demonstrated the highest accuracy and fastest results for object detection.
- Automatic assignment of gaze data to objects significantly simplifies data analysis.
- This approach enables potential real-time system responses based on user gaze.
Conclusions:
- Object recognition models, particularly YOLOv4, can effectively automate mobile eye-tracking data analysis in educational settings.
- This automation addresses the limitations of manual analysis, making larger studies feasible.
- Several challenges in applying object detection to mobile eye-tracking data were identified and discussed.
