Research map / Vision-language and multimodal / Visual question answering
Invariant Grounding for Video Question Answering
Check 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 |
|---|---|---|---|
| Train with IGV instead of the strongest listed baseline | VideoQA models · NExT-QA test, All | accuracy: 50.01 -> 51.34 | Helped |
| Train with IGV instead of the strongest listed baseline | VideoQA models · MSVD-QA test | accuracy: 39.4 -> 40.8 | Helped |
| Train with IGV instead of the strongest listed baseline | VideoQA models · MSRVTT-QA test | accuracy: 36.9 -> 38.3 | Helped |
| Add IGV training to Co-Mem | Co-Mem · MSVD-QA test | accuracy: 34.6 -> 37.7 | Helped |
| Add IGV training to HGA | HGA · MSVD-QA test | accuracy: 35.4 -> 38.8 | Helped |
| Add IGV training to Co-Mem | Co-Mem · MSRVTT-QA test | accuracy: 35.3 -> 37.3 | Helped |
| Add IGV training to HGA | HGA · MSRVTT-QA test | accuracy: 36.1 -> 37.5 | Helped |
| Use only causal-prediction loss rather than the full IGV objective (ablation) | Our Backbone · MSVD-QA | accuracy: 40.8 -> 36.0 | Hurt |
| Use only causal-prediction loss rather than the full IGV objective (ablation) | Co-Mem · MSVD-QA | accuracy: 37.7 -> 33.3 | Hurt |
| Add complement-prediction and intervened-prediction losses to causal-prediction loss (ablation) | Our Backbone · MSVD-QA | accuracy: 36.0 -> 40.8 | Helped |
| Add complement-prediction and intervened-prediction losses to causal-prediction loss (ablation) | Our Backbone · MSRVTT-QA | accuracy: 36.7 -> 38.3 | Helped |
Closest papers by meaning
- Cross-modal Causal Relation Alignment for Video Question Grounding (CVPR 2025 · similarity 0.91)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.91)
- Commonsense Video Question Answering through Video-Grounded Entailment Tree Reasoning (CVPR 2025 · similarity 0.90)
- VTQA: Visual Text Question Answering via Entity Alignment and Cross-Media Reasoning (CVPR 2024 · similarity 0.90)
- Can I Trust Your Answer? Visually Grounded Video Question Answering (CVPR 2024 · similarity 0.90)
- MoReVQA: Exploring Modular Reasoning Models for Video Question Answering (CVPR 2024 · 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.