GazePrompt: Enhancing Low Vision People's Reading Experience with Gaze-Aware Augmentations

Title of the Paper

GazePrompt: Enhancing Low Vision People’s Reading Experience with Gaze-Aware Augmentations

Paper Information

• Research Domain: Human-Computer Interaction and Assistive Technology, specifically technologies to enhance the reading experience for people with low vision
• Keywords: Accessibility, Low Vision, Eye Tracking, Visual Augmentation, Reading Assistance

Research Background and Problem Statement

• What problems or challenges did the authors identify?

  • Major challenges faced by people with low vision during reading include:
    1. Locating the next line: Difficulty in quickly locating the next line of text when using full-screen magnification tools.
    2. Identifying difficult words: Visual impairments make it harder to recognize complex or visually similar words.
  • Although existing assistive tools for low vision (e.g., magnifiers, screen magnification features) provide some level of support, they still have significant limitations.

• Why is this problem important?

  • Reading is a critical means of acquiring information in daily life, yet the reading experience for people with low vision is significantly inferior to that of individuals with normal vision.
  • According to the World Health Organization, at least 2.2 billion people globally are affected by vision impairment, with this number expected to double in the next decade. Thus, improving the reading ability of people with low vision is of global importance and urgency.

• Research Motivation and Related Work

  • Existing studies show that eye-tracking technology can enable personalized assistive functions by detecting users’ low-level eye movement behaviors. However, assistive technologies leveraging eye tracking for people with low vision are still in their early stages, with most research focused on visual science rather than practical tools for daily use.
  • Current tool designs and evaluations predominantly target individuals with normal vision, failing to adequately address the needs of the low vision population.

Solution

• What methods or solutions did the authors propose?

  • The authors proposed an eye-tracking-based reading assistance system, GazePrompt, which provides two types of augmentations:
    1. Line-Switching Support: Identifies the line the user is reading or planning to read and enhances it visually in real time.
    2. Difficult-Word Support: Detects when users spend excessive time on a specific word and provides visual magnification or text-to-speech assistance for that word.

• What are the innovative aspects of this solution?

  • Targeted Augmentation: Unlike traditional full-screen magnification methods, eye-tracking augmentation can identify user targets in real time and provide focused enhancements.
  • Diverse Design Options: Offers two design choices (e.g., line highlighting and arrows, magnification and text-to-speech) to accommodate varying preferences and visual conditions among low vision users.
  • Dynamic Interaction: Combines real-time eye movement monitoring with user behavior, reducing the learning curve and operational burden.

• Implementation Steps and Key Technologies:

  • Line-Switching Augmentation:
    • Uses eye-tracking data to locate the user’s current and target lines, employing a weighted voting mechanism to identify the reading target line.
    • Provides two enhancement options: “line highlighting” and “arrows” for directional guidance.
  • Difficult-Word Augmentation:
    • Detects difficult words based on initial and total fixation duration, as well as regression behaviors.
    • Offers additional assistance via “magnification display” or “text-to-speech” methods.
  • System Architecture:
    • Implements customized calibration methods to improve eye-tracking data accuracy, including point calibration and line calibration.
    • Uses Tobii Pro SDK to capture user eye-tracking data and processes it with real-time algorithms.

Research Outcomes

• What specific results were achieved?

  • In reading tests, GazePrompt significantly reduced participants’ line-switching time and improved flexibility during page navigation.
  • Reduced the total number of misidentified words, enhancing perceived reading focus and comprehension.

• How does it compare to existing solutions?

  • Faster and more accurate visual assistance functionality.
  • Provides personalized enhancements directly supporting user behavior.
  • Effectively avoids the operational burden and cognitive load associated with traditional methods.

• What were the experimental or evaluation results?

  • Two experiments (reading aloud and silent reading):
    • Reading Aloud Experiment (13 low vision participants):
      • Significantly reduced line-switching time.
      • Although line-switching accuracy did not show significant improvement, user feedback indicated a noticeable enhancement in reading experience.
    • Silent Reading Experiment (another 13 low vision participants):
      • Encouraged participants to focus more on content comprehension, with difficult-word augmentation serving as a word confirmation tool.
  • Qualitative surveys revealed clear user preferences for the augmentation designs, such as their applicability in technical texts and low-light environments.

• Limitations and Future Directions:

  • Sample Limitations: The number of low vision participants in the experiments was limited; future studies should expand the sample to cover diverse visual conditions.
  • Calibration Challenges: The current calibration process poses difficulties for individuals with severe vision loss; future work should develop more accessible calibration methods.
  • Interaction Optimization: Consider combining manual control with eye-tracking control to enhance usability and accuracy.
  • Personalization Improvements: Explore adaptive parameter adjustments to accommodate individual reading habits.

This paper boldly explores the practical application of eye-tracking technology in the field of low vision assistance, demonstrating its potential for broad impact while highlighting areas for further optimization and development.