核心概念
AI-powered scene description applications offer valuable daily tools for blind and low vision individuals, but improvements are needed for user satisfaction and trust.
摘要
In this study, researchers conducted a two-week diary study with 16 blind and low vision participants to explore the use cases of an AI-powered scene description application. The participants shared their information goals and assessments of the visual descriptions they received. Common goals included identifying objects, understanding surroundings, and reading text. Participants also tested the application against their knowledge and avoided touching inconvenient or dangerous objects. Overall, users rated descriptions relatively low on satisfaction and trust, indicating room for improvement in delivering satisfactory experiences.
Structure:
Introduction to challenges faced by blind and low vision individuals.
Overview of existing research on visual interpretation applications.
Study design: Two-week diary study with 16 participants using an AI-powered scene description application.
Analysis of user goals: Identifying specific objectives such as identifying subjects or features.
Examination of photo content captured by participants: Focus on stationary subjects like decor, living subjects, and building architecture.
Unique use cases: Testing the application against current knowledge, determining subject location or presence, informing camera use.
Inspection of inconvenient objects through photos to avoid touching dirty or dangerous items.
Discussion on the importance of scene description applications in enhancing daily experiences for blind and low vision users.
統計資料
We found that users scored the descriptions relatively low on average: 2.76 out of 5 for satisfaction and 2.43 out of 4 for trust.
Participants submitted an average of 22 entries each during the two-week diary study.