Research map / Vision-language and multimodal / Visual question answering
Cross-Modal Causal Relational Reasoning for Event-Level Visual Question Answering
Yang Liu, Guanbin Li, Liang Lin
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 the hierarchical semantic-role parser (HSRP) (ablation) | CMCIR · SUTD-TrafficQA | accuracy: 37.65 -> 38.58 | Improved |
| add linguistic back-door causal intervention (ablation) | CMCIR · SUTD-TrafficQA | accuracy: 37.71 -> 38.58 | Improved |
| add visual front-door causal intervention (ablation) | CMCIR · SUTD-TrafficQA | accuracy: 37.68 -> 38.58 | Improved |
| add the causality-aware visual-linguistic reasoning module (CVLR) (ablation) | CMCIR · SUTD-TrafficQA | accuracy: 37.42 -> 38.58 | Improved |
| add semantic graph embedding (ablation) | CMCIR · SUTD-TrafficQA | accuracy: 37.93 -> 38.58 | Improved |
| add adaptive linguistic feature fusion (ablation) | CMCIR · SUTD-TrafficQA | accuracy: 37.84 -> 38.58 | Improved |
| add CVLR (ablation) | CMCIR · MSVD-QA | accuracy: 42.0 -> 43.7 | Improved |
| add CVLR (ablation) | CMCIR · MSRVTT-QA | accuracy: 38.0 -> 38.9 | Improved |
| add HSRP (ablation) | CMCIR · TGIF-QA (Action) | accuracy: 75.4 -> 78.1 | Improved |
| use 8 MMA heads instead of 1 | CMCIR · SUTD-TrafficQA | accuracy: 37.83 -> 38.58 | Improved |
| use 1 GCN layer instead of 2 | CMCIR · SUTD-TrafficQA | accuracy: 37.84 -> 38.58 | Improved |
| use more MTB layers (3 vs 8) | CMCIR · SUTD-TrafficQA | accuracy: 38.58 -> 37.58 | 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.
- HAIR — HAIR: Hierarchical Visual-Semantic Relational Reasoning for Video Question Answering (ICCV 2021) · better on 1 of 1 shared result
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 0.90)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.89)
- From Prompts to Tokens: Internalizing Causal Supervision in Vision-Language Model for Multi-Image Causal Reasoning (ICML 2026 · similarity 0.89)
- Visual Chain-of-Thought Prompting for Knowledge-Based Visual Reasoning (AAAI 2024 · similarity 0.88)
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · similarity 0.87)
- The Abstraction Gap in Vision-Language Causal Reasoning (ICML 2026 · similarity 0.87)
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.
