Research map / Video / Video-language understanding
VUDG: A Dataset for Video Understanding Domain Generalization
Ziyi Wang, Zhi Gao, Boxuan Yu, Zirui Dai, Peiyao Wang, Yuxiang Song, Qingyuan Lu, Jin Chen, Xinxiao Wu
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 |
|---|---|---|---|
| use LLM-based video models instead of non-LLM VideoQA methods | multiple domain generalization on VUDG · VUDG multiple DG | accuracy: 17.3 -> 66.9 | Improved |
| move from multiple DG to single DG training | VideoLLaMA2-7B · VUDG Env. Condition | accuracy: degrades 4.7% | Hurt |
| fine-tune with multiple DG instead of using zero-shot | Qwen2.5VL-3B · VUDG D-Avg | accuracy: 69.2 -> 61.4 | Hurt |
| fine-tune with multiple DG instead of using zero-shot | VideoLLaMA2-7B · VUDG D-Avg | accuracy: 33.4 -> 66.9 | Improved |
| scale Qwen2.5VL from 3B to 7B | Qwen2.5VL zero-shot · VUDG multiple-choice D-Avg | accuracy: 69.2 -> 72.1 | Improved |
| ask temporal questions instead of object identification questions | Qwen2.5VL-7B · VUDG multiple-choice by category | accuracy: 80.8 -> 67.7 | Hurt |
| use open-ended instead of multiple-choice evaluation | zero-shot LVLMs · VUDG | model gap | Mixed |
Closest papers by meaning
- VideoDG: Generalizing Temporal Relations in Videos to Novel Domains (TPAMI 2021 · similarity 0.90)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.89)
- Transformer-Empowered Invariant Grounding for Video Question Answering (TPAMI 2023 · similarity 0.89)
- Sink-Token-Aware Pruning for Fine-Grained Video Understanding in Efficient Video LLMs (ECCV 2026 · similarity 0.89)
- CG-Bench: Clue-grounded Question Answering Benchmark for Long Video Understanding (ICLR 2025 · similarity 0.88)
- Leveraging Textual Memory and Key Frame Reasoning for Full Video Understanding Using Off-the-Shelf LLMs and VLMs (Student Abstract) (AAAI 2025 · 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.
