Research map / Video / Video-language understanding
Video Graph Transformer for Video Question Answering
Junbin Xiao, Pan Zhou, Tat-Seng Chua, Shuicheng Yan
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 dynamic graph transformer module (ablation) | VGT · NExT-QA val | Acc@All: 53.22 -> 55.02 | Improved |
| add the dynamic graph transformer module (ablation) | VGT · TGIF-QA Action | accuracy: 89.6 -> 95.0 | Improved |
| add temporal graph transformer (node and edge) (ablation) | VGT · NExT-QA val | Acc@All: 53.74 -> 55.02 | Improved |
| add node transformer (ablation) | VGT · NExT-QA val | Acc@All: 53.84 -> 55.02 | Improved |
| add edge transformer (ablation) | VGT · NExT-QA val | Acc@All: 54.30 -> 55.02 | Improved |
| add global frame feature F_I (ablation) | VGT · NExT-QA val | Acc@All: 53.58 -> 55.02 | Improved |
| use similarity comparison instead of answer classification (ablation) | VGT · NExT-QA val | Acc@All: 45.82 -> 55.02 | Improved |
| use similarity comparison instead of answer classification (ablation) | VGT · FrameQA | accuracy: 56.9 -> 61.6 | Improved |
| replace cross-modal transformer with light cross-modal interaction in classification variant (ablation) | Comp->CLS · NExT-QA val | Acc@All: 45.82 -> 46.98 | No clear effect |
| replace S3D with DGT encoder in VQA-T (ablation) | VQA-T · NExT-QA val | Acc@All: 45.30 -> 50.02 | Improved |
| replace cross-modal transformer with cross-modal interaction module in VQA-T (ablation) | VQA-T · NExT-QA val | Acc@All: 45.30 -> 45.40 | No clear effect |
| add cross-modal interaction (ablation) | VGT · TGIF-QA | accuracy: more than 10% | Improved |
Closest papers by meaning
- Contrastive Video Question Answering via Video Graph Transformer (TPAMI 2023 · similarity 0.95)
- MIST: Multi-Modal Iterative Spatial-Temporal Transformer for Long-Form Video Question Answering (CVPR 2023 · similarity 0.91)
- Video-Context Aligned Transformer for Video Question Answering (AAAI 2024 · similarity 0.89)
- Commonsense Video Question Answering through Video-Grounded Entailment Tree Reasoning (CVPR 2025 · similarity 0.89)
- VideoEspresso: A Large-Scale Chain-of-Thought Dataset for Fine-Grained Video Reasoning via Core Frame Selection (CVPR 2025 · similarity 0.89)
- MoReVQA: Exploring Modular Reasoning Models for Video Question Answering (CVPR 2024 · similarity 0.89)
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.
