Research map / Vision-language and multimodal / Visual question answering
Human-Adversarial Visual Question Answering
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Fine-tune VisualBERT further on AdVQA val (ablation) | VisualBERT · AdVQA test | accuracy: 28.70 to 33.10 | Helped |
| Fine-tune ViLBERT further on AdVQA val (ablation) | ViLBERT · AdVQA test | accuracy: 27.36 to 32.85 | Helped |
| Use an OCR-capable M4C model trained on VQA v2 instead of TextVQA+STVQA | M4C · AdVQA test | accuracy: 28.86±0.35 to 33.52±0.47 | Helped |
| Add image input to a question-only BERT baseline | VQA models · AdVQA test | accuracy: 26.90±0.36; multimodal scores range from 25.14±0.93 to 33.52±0.47 | Mixed |
| Evaluate VQA systems on human-adversarial rather than VQA v2 questions | VQA models · AdVQA test vs VQA v2 test-dev | accuracy: VisualBERT: 70.37±0.05 to 28.70±0.36 | Hurt |
Closest papers by meaning
- Tips and Tricks for Visual Question Answering: Learnings From the 2017 Challenge (CVPR 2018 · similarity 0.93)
- VQAttack: Transferable Adversarial Attacks on Visual Question Answering via Pre-trained Models (AAAI 2024 · similarity 0.92)
- NaturalBench: Evaluating Vision-Language Models on Natural Adversarial Samples (NeurIPS 2024 · similarity 0.91)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.90)
- FRAMES-VQA: Benchmarking Fine-Tuning Robustness across Multi-Modal Shifts in Visual Question Answering (CVPR 2025 · similarity 0.90)
- Mind the Uncertainty in Human Disagreement: Evaluating Discrepancies Between Model Predictions and Human Responses in VQA (AAAI 2025 · similarity 0.90)
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.