AI Paper Atlas

Research map / Vision-language and multimodal / Visual question answering

Identifying and Resolving Pitfalls of Knowledge-Based VQA Benchmarks: Auditing, Repairing, and Augmenting

Qian Ma, S M Rayeed, Qiong Wu, Charles Stewart, Yao Ma

ECCV 2026 · Vision-language and multimodal › Visual question answering · approach: Universal VQA models · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
repair answers and questions (audit-and-repair protocol)IBA · InfoSeek AllVQA accuracy: 34.5 -> 42.4Improved
repair InfoSeek annotationsWiki-PRF · InfoSeek TimeVQA accuracy: 35.1 -> 43.3Improved
repair InfoSeek annotationsWiki-PRF · InfoSeek AllVQA accuracy: 44.9 -> 43.6Hurt
repair E-VQA questionsIBA · E-VQABEM accuracy: 41.9 -> 42.7No clear effect
fix benchmark: ranking of IBA vs ReflectiVAIBA vs ReflectiVA · InfoSeek AllVQA accuracy: ReflectiVA 37.3 vs IBA 34.5 before; IBA 42.4 vs ReflectiVA 38.1 afterMixed
question-only fixes on grounded InfoSeek subset (ablation)EchoSight · InfoSeek shared grounded subset (664)QA accuracy: 60.1 -> 75.1Improved
answer-only fixes on grounded InfoSeek subset (ablation)EchoSight · InfoSeek shared grounded subset (664)QA accuracy: 44.2 -> 44.3No clear effect
add single distractor entity (intra-category)EVA-CLIP-8B retrieval · InfoSeekRecall@1: 43.5 -> 14.7Hurt
add single distractor entity (intra-category)EVA-CLIP-8B retrieval · E-VQARecall@1: 13.4 -> 3.5Hurt
add intra-category distractorIBA · InfoSeekVQA accuracy: 40.1 -> 21.4Hurt
add intra-category distractorWiki-PRF · InfoSeekVQA accuracy: 43.9 -> 23.6Hurt
post-retrieval evidence selection under augmentationIBA · InfoSeek post-retrieval recallrecall: 45.1 -> 21.7Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy