DARE: Diverse Visual Question Answering with Robustness Evaluation

2024-09-26

arXiv.org

AI Summary

Current vision-language models (VLMs) exhibit poor performance on critical vision-language (VL) reasoning tasks—such as counting and spatial reasoning—and lack robustness to minor perturbations in prompts, answer option subsets, output formats, or answer cardinality. To address this, we propose DARE, a multi-dimensional, robustness-driven visual question answering benchmark covering five challenging VL task categories. DARE uniquely integrates diverse capability evaluation with four-dimensional robustness assessment—across prompt, option, format, and answer-space variations. It employs human-verified data, adversarial option perturbations, format/semantic consistency constraints, and a cross-model standardization framework. Experiments reveal that state-of-the-art models achieve below 40% accuracy on several tasks; performance degrades by up to 34% under worst-case option subsets; open-source models (e.g., LLaVA-1.6, Idefics2) demonstrate significantly lower robustness than proprietary models (GPT-4, Gemini), though the latter remain highly fragile. DARE is publicly released.

Technology Category

• Computer Vision: Large Vision Models
• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Vision Language Models (VLMs) extend remarkable capabilities of text-only large language models and vision-only models, and are able to learn from and process multi-modal vision-text input. While modern VLMs perform well on a number of standard image classification and image-text matching tasks, they still struggle with a number of crucial vision-language (VL) reasoning abilities such as counting and spatial reasoning. Moreover, while they might be very brittle to small variations in instructions and/or evaluation protocols, existing benchmarks fail to evaluate their robustness (or rather the lack of it). In order to couple challenging VL scenarios with comprehensive robustness evaluation, we introduce DARE, Diverse Visual Question Answering with Robustness Evaluation, a carefully created and curated multiple-choice VQA benchmark. DARE evaluates VLM performance on five diverse categories and includes four robustness-oriented evaluations based on the variations of: prompts, the subsets of answer options, the output format and the number of correct answers. Among a spectrum of other findings, we report that state-of-the-art VLMs still struggle with questions in most categories and are unable to consistently deliver their peak performance across the tested robustness evaluations. The worst case performance across the subsets of options is up to 34% below the performance in the standard case. The robustness of the open-source VLMs such as LLaVA 1.6 and Idefics2 cannot match the closed-source models such as GPT-4 and Gemini, but even the latter remain very brittle to different variations.

Problem

• Evaluating VLMs' robustness in diverse vision-language reasoning tasks
• Assessing performance variations under different evaluation protocols
• Comparing open-source and closed-source VLMs' robustness and brittleness

Innovation

• Introduces DARE benchmark for VLM evaluation
• Evaluates five diverse VQA categories robustly
• Tests robustness via prompt and format variations