Research map / Video / Video-language understanding
Wavelet-based Intra-video Counterfactual Reasoning for Video Question Grounding
Jiasheng Yuan, Wei Wei
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 |
|---|---|---|---|
| add wavelet-based dynamic modeling (WDM) (ablation) | WICI · NeXT-GQA | Acc@GQA: 17.8 -> 19.0 | Improved |
| add intra-video counterfactual mining (ICM) (ablation) | WICI · NeXT-GQA | Acc@GQA: 17.7 -> 19.0 | Improved |
| add shuffled-video counterfactual samples (ablation) | WICI ICM · NeXT-GQA | Acc@GQA: 18.2 -> 19.0 | Improved |
| use curriculum-guided counterfactual mining instead of fixed distance (ablation) | WICI ICM · NeXT-GQA | Acc@GQA: 18.2 -> 19.0 | Improved |
| use margin-based loss instead of contrastive loss (ablation) | WICI ICM · NeXT-GQA | Acc@GQA: 18.3 -> 19.0 | Improved |
| use more than two wavelet decomposition levels (ablation) | WDM · NeXT-GQA | Acc@GQA | Hurt |
| apply WICI to Temp[CLIP] backbone versus NG+ | WICI · NeXT-GQA | Acc@GQA: 16.0 -> 19.0 | Improved |
| apply WICI versus CRA | WICI · NeXT-GQA | Acc@GQA: 18.2 -> 19.0 | Improved |
| apply WICI versus CRA on STAR | WICI · STAR | Acc@GQA: 26.8 -> 29.2 | Improved |
| WICI versus TimeCraft and CRA on IoU-based grounding | WICI · NeXT-GQA | TIoU@0.5: 10.6 (CRA) vs 10.2 | Mixed |
| add WICI and shuffle frames at inference | WICI · NeXT-GQA | Acc@GQA: 19.0 -> 15.0 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TimeCraft — TimeCraft: Navigate Weakly-Supervised Temporal Grounded Video Question Answering via Bi-directional Reasoning (ECCV 2024) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Cross-modal Causal Relation Alignment for Video Question Grounding (CVPR 2025 · similarity 0.91)
- Two Causally Related Needles in a Video Haystack (NeurIPS 2025 · similarity 0.89)
- Commonsense Video Question Answering through Video-Grounded Entailment Tree Reasoning (CVPR 2025 · similarity 0.88)
- Temporal Chain of Thought: Long-Video Understanding by Thinking in Frames (NeurIPS 2025 · similarity 0.88)
- VideoEspresso: A Large-Scale Chain-of-Thought Dataset for Fine-Grained Video Reasoning via Core Frame Selection (CVPR 2025 · similarity 0.88)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.88)
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.
