Grounding Answers for Visual Questions Asked by Visually Impaired People

TL;DR

This paper introduces the VizWiz-VQA-Grounding dataset, designed to evaluate how well models can locate visual evidence in images for questions asked by visually impaired users, revealing current models' limitations.

Contribution

The paper presents the first dataset specifically for grounding answers in visual questions from visually impaired users and analyzes the performance gaps of existing models.

Findings

Current SOTA models often fail to identify correct visual evidence.

Models struggle with small visual evidence and high-quality images.

Text recognition questions pose additional challenges.

Abstract

Visual question answering is the task of answering questions about images. We introduce the VizWiz-VQA-Grounding dataset, the first dataset that visually grounds answers to visual questions asked by people with visual impairments. We analyze our dataset and compare it with five VQA-Grounding datasets to demonstrate what makes it similar and different. We then evaluate the SOTA VQA and VQA-Grounding models and demonstrate that current SOTA algorithms often fail to identify the correct visual evidence where the answer is located. These models regularly struggle when the visual evidence occupies a small fraction of the image, for images that are higher quality, as well as for visual questions that require skills in text recognition. The dataset, evaluation server, and leaderboard all can be found at the following link: https://vizwiz.org/tasks-and-datasets/answer-grounding-for-vqa/.