Tell Me the Evidence? Dual Visual-Linguistic Interaction for Answer Grounding

TL;DR

This paper introduces DaVI, a dual visual-linguistic framework for answer grounding in VQA that enhances interpretability and flexibility by integrating visual and linguistic interactions, achieving top performance in a major challenge.

Contribution

DaVI is the first unified end-to-end model enabling both answer generation and visual evidence grounding through dual interaction mechanisms.

Findings

Ranked 1st in 2022 VizWiz Grand Challenge answer grounding

Outperforms previous methods in visual question answering interpretability

Demonstrates effective visual-linguistic integration

Abstract

Answer grounding aims to reveal the visual evidence for visual question answering (VQA), which entails highlighting relevant positions in the image when answering questions about images. Previous attempts typically tackle this problem using pretrained object detectors, but without the flexibility for objects not in the predefined vocabulary. However, these black-box methods solely concentrate on the linguistic generation, ignoring the visual interpretability. In this paper, we propose Dual Visual-Linguistic Interaction (DaVI), a novel unified end-to-end framework with the capability for both linguistic answering and visual grounding. DaVI innovatively introduces two visual-linguistic interaction mechanisms: 1) visual-based linguistic encoder that understands questions incorporated with visual features and produces linguistic-oriented evidence for further answer decoding, and 2) linguistic-based visual decoder that focuses visual features on the evidence-related regions for answer grounding. This way, our approach ranked the 1st place in the answer grounding track of 2022 VizWiz Grand Challenge.