Research map / Video / Video-language understanding
IntentQA: Intent Question Answering in Videos by Cognitive Context Reasoning
J H Li, Ping Wei, Wenjuan Han, Song Chun Zhu, Lifeng Fan
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 Video Query Language (VQL) (ablation) | Base Model · IntentQA test | accuracy (%): 51.36 -> 52.34 (+0.98) | Improved |
| add triplet loader (ablation) | Base + VQL · IntentQA test | accuracy (%): 52.34 -> 53.80 (+1.46) | Improved |
| add triplet margin loss (ablation) | Base + VQL + loader · IntentQA test | accuracy (%): 53.80 -> 54.50 (+0.70) | Improved |
| add LLM commonsense with confidence prompting (ablation) | visual model · IntentQA test | accuracy (%): 54.50 -> 58.48 (+3.98) | Improved |
| use confidence prompting instead of late fusion (ablation) | InstructGPT · IntentQA test | accuracy (%): 57.64 -> 58.48 | Improved |
| add caption information (ablation) | X-CaVIR (InstructGPT) · IntentQA test | accuracy (%): 58.48 -> 60.73 (+2.25) | Improved |
| add caption information (ablation) | X-CaVIR (GPT-4) · IntentQA test | accuracy (%): 62.65 -> 64.81 | Improved |
| add the visual model on top of caption + LLM (ablation) | InstructGPT · IntentQA test | accuracy (%): 54.03 -> 60.73 | Improved |
| use GPT-4 instead of InstructGPT | X-CaVIR full · IntentQA test | accuracy (%): 60.73 -> 64.81 | Improved |
| use random masking instead of attention-based masking in contrastive learning (ablation) | contrastive learning · IntentQA test | accuracy (%): 54.50 -> 51.50 | Hurt |
| mask lowest-attention tokens instead of selected action (ablation) | contrastive learning · IntentQA test | accuracy (%): 54.50 -> 54.05 | Hurt |
| compare to the strongest prior VideoQA model | X-CaVIR vs VGT · IntentQA test | accuracy (%): 51.27 -> 60.73 | Improved |
Closest papers by meaning
- IntentQA: Context-aware Video Intent Reasoning (ICCV 2023 · similarity 0.92)
- BDIQA: A New Dataset for Video Question Answering to Explore Cognitive Reasoning through Theory of Mind (AAAI 2024 · similarity 0.89)
- Black Swan: Abductive and Defeasible Video Reasoning in Unpredictable Events (CVPR 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)
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · similarity 0.87)
- Temporal Chain of Thought: Long-Video Understanding by Thinking in Frames (NeurIPS 2025 · 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.
