VUDG: A Dataset for Video Understanding Domain Generalization

TL;DR

VUDG introduces a comprehensive dataset to evaluate the robustness of video understanding models across diverse domain shifts, revealing current models' vulnerabilities and guiding future research in domain generalization.

Contribution

The paper presents VUDG, a new dataset with videos from 11 domains for assessing domain generalization in video understanding, and analyzes model performance under domain shifts.

Findings

Most models degrade in performance under domain shifts.

State-of-the-art LVLMs show significant robustness gaps.

VUDG serves as a benchmark for future domain generalization research.

Abstract

Video understanding has made remarkable progress in recent years, largely driven by advances in deep models and the availability of large-scale annotated datasets. However, existing works typically ignore the inherent domain shifts encountered in real-world video applications, leaving domain generalization (DG) in video understanding underexplored. Hence, we propose Video Understanding Domain Generalization (VUDG), a novel dataset designed specifically for evaluating the DG performance in video understanding. VUDG contains videos from 11 distinct domains that cover three types of domain shifts, and maintains semantic similarity across different domains to ensure fair and meaningful evaluation. We propose a multi-expert progressive annotation framework to annotate each video with both multiple-choice and open-ended question-answer pairs. Extensive experiments on 9 representative large video-language models (LVLMs) and several traditional video question answering methods show that most models (including state-of-the-art LVLMs) suffer performance degradation under domain shifts. These results highlight the challenges posed by VUDG and the difference in the robustness of current models to data distribution shifts. We believe VUDG provides a valuable resource for prompting future research in domain generalization video understanding.