AI Paper Atlas

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

Robust Visual Question Answering: Datasets, Methods, and Future Challenges

Jie Ma, Pinghui Wang, Dechen Kong, Zewei Wang, Jun Liu, Hongbin Pei, Junzhou Zhao

TPAMI 2024 · Vision-language and multimodal › Visual question answering · approach: Universal VQA models · 64 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
use pretrained LXMERT backbone with debiasing instead of UpDnLoss-Rescaling · VQA-CP v2 testaccuracy: 47.09 -> 66.40Improved
add LMH ensemble debiasing to UpDnUpDn · VQA v2 validationaccuracy: 65.99 -> 55.99Hurt
add CSS contrastive learning (CSS+CL) on LMHLMH · VQA-CP v2 testaccuracy: 58.95 -> 59.18Mixed
generative ALBEF (4M) instead of discriminativeALBEF · VQA-CP v2 testaccuracy: 44.40 -> 49.20Improved
scale ALBEF pretraining data from 4M to 14MALBEF DISC · VQA-CP v2 testaccuracy: 44.40 -> 45.20Improved
add large-scale pretraining to ALBEF GENALBEF GEN · VQA-CP v2 testaccuracy: 36.60 -> 49.20Improved
move from in-distribution to OOD splitUNITER Base · VQA v2 val vs VQA-CP v2accuracy: 72.20 -> 46.93Hurt
use UNITER Large instead of BaseUNITER · VQA-CP v2 testaccuracy: 46.93 -> 50.98Improved
use SAR answer re-ranking on LMHLMH · VQA-CP v2 testaccuracy: 52.87 -> 66.73Improved

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