A Visual Attentive Model for Discovering Patterns in Eye-Tracking Data-A Proposal in Cultural Heritage.

Area of Science:

• Computer Science
• Human-Computer Interaction
• Cultural Heritage Technology

Background:

• Museums utilize technology to personalize visitor experiences.
• Determining visitor interest is a significant challenge in cultural heritage settings.
• Eye-tracking data offers insights into visual attention and engagement.

Purpose of the Study:

• To propose a novel Visual Attentive Model (VAM) for identifying visitor interest based on eye-tracking data.
• To differentiate between adult and child visitors using their distinctive eye movement patterns.
• To enhance the personalization of museum visits in the cultural heritage context.

Main Methods:

• Collected eye-tracking data from adults and children observing specific artworks with analogous features.
• Developed a novel coordinate representation for eye sequences using Geometric Algebra.
• Integrated this representation with a deep learning model (Deep Convolutional Neural Networks - DCNNs) for automated recognition.

Main Results:

• The Visual Attentive Model (VAM) achieved high accuracy (over 80%) in identifying adults and children.
• Demonstrated the effectiveness of combining Geometric Algebra with DCNNs for analyzing eye movement patterns.
• Validated the approach on a curated set of paintings with similar visual characteristics.

Conclusions:

• The proposed Visual Attentive Model (VAM) is effective for identifying visitor demographics (adults vs. children) in museums.
• This technology can significantly improve the personalization of cultural heritage experiences.
• Eye movement pattern analysis holds potential for understanding visitor engagement and preferences.