AI Paper Atlas

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

ICLR 2026 · Video › Video-language understanding · approach: Long video understanding · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
use LLM-based video models instead of non-LLM VideoQA methodsmultiple domain generalization on VUDG · VUDG multiple DGaccuracy: 17.3 -> 66.9Improved
move from multiple DG to single DG trainingVideoLLaMA2-7B · VUDG Env. Conditionaccuracy: degrades 4.7%Hurt
fine-tune with multiple DG instead of using zero-shotQwen2.5VL-3B · VUDG D-Avgaccuracy: 69.2 -> 61.4Hurt
fine-tune with multiple DG instead of using zero-shotVideoLLaMA2-7B · VUDG D-Avgaccuracy: 33.4 -> 66.9Improved
scale Qwen2.5VL from 3B to 7BQwen2.5VL zero-shot · VUDG multiple-choice D-Avgaccuracy: 69.2 -> 72.1Improved
ask temporal questions instead of object identification questionsQwen2.5VL-7B · VUDG multiple-choice by categoryaccuracy: 80.8 -> 67.7Hurt
use open-ended instead of multiple-choice evaluationzero-shot LVLMs · VUDGmodel gapMixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy