FURI | Spring 2026

CustomQA: Customized Video Descriptions and Video Question Answering for Blind and Low Vision Users

Data icon, disabled. Four grey bars arranged like a vertical bar chart.

Traditionally, Blind and Low Vision (BLV) users consume videos through Audio Descriptions (ADs), visual descriptions written and narrated by sighted describers. However, these descriptions struggle to cover individualized BLV needs.

This project introduces CustomQA, a Chrome extension allowing BLV users to generate customizable ADs and ask visual questions about videos. CustomQA provides AD customization for frequency, emphasis, color, subjectivity, and control over Visual Question Answer (VQA) response length. Additionally, users can adjust volume, gender, and speed of AI text-to-speech.

User studies are performed to discover common question types and questioning frequencies, including effects customization options may have on questioning behaviors.

Student researcher

Nghi Huynh

Computer science

Hometown: Tempe, AZ, United States

Graduation date: Spring 2026