Research map / Vision-language and multimodal / Visual question answering
Toward Unsupervised Realistic Visual Question Answering
Yuwei Zhang, Chih-Hui Ho, Nuno Vasconcelos
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 changed | Where | Measured | Effect |
|---|---|---|---|
| train with pseudo UQs from randomly paired images and questions | BUTD/UNITER/LXMERT · RGQA | AUAF: 36.50 -> 47.02 (BUTD, MSP vs RP) | Improved |
| add RoI Mixup to pseudo UQ training (ablation) | RP · RGQA | AUAF: 47.02 -> 47.60 (BUTD) | Improved |
| ensemble RP and Mix models (ablation) | Mix · RGQA | AUAF: 47.60 -> 48.28 (BUTD) | Improved |
| use post-hoc ODIN, Maha or Energy instead of MSP | VQA classifiers · RGQA | AUAF: 36.50 -> 36.50 (BUTD, ODIN) | No clear effect |
| train only on hard pseudo UQs instead of random pseudo UQs (ablation) | RP · RGQA | AUAF: 47.02 -> 40.2 (BUTD) | Hurt |
| replace random pseudo UQ sampling with Resample adversarial weighting (ablation) | RP · RGQA | AUAF: 47.02 -> 38.07 (BUTD) | Hurt |
| use larger and more recent VQA models | MSP RVQA · RGQA | AUAF: 36.50 (BUTD) -> 44.89 (VinVL) | No clear effect |
| use integrated architecture instead of separated or branched (ablation) | RP with LXMERT · RGQA | AUAF | Improved |
| train with pseudo UQs | LXMERT · GQA test | accuracy: 77.8 -> 76.3 (RP), 77.1 (Mix) | No clear effect |
Closest papers by meaning
- UNK-VQA: A Dataset and a Probe Into the Abstention Ability of Multi-Modal Large Models (TPAMI 2024 · similarity 0.93)
- Don't Just Assume; Look and Answer: Overcoming Priors for Visual Question Answering (CVPR 2018 · similarity 0.92)
- Answer Them All! Toward Universal Visual Question Answering Models (CVPR 2019 · similarity 0.92)
- FVQA: Fact-Based Visual Question Answering (TPAMI 2017 · similarity 0.92)
- Roses Are Red, Violets Are Blue... but Should VQA Expect Them To? (CVPR 2021 · similarity 0.92)
- Robust Visual Question Answering: Datasets, Methods, and Future Challenges (TPAMI 2024 · similarity 0.92)
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.
