AnyGroundBench：面向视觉-语言模型中视频定位的专业领域基准

摘要

视觉-语言模型（VLM）在时空视频定位（STVG）任务中展现出巨大潜力。然而，当前评估方案主要局限于通用日常场景基准的零样本测试，这与专业领域的实际应用存在严重脱节——在这些场景中，模型不可避免地会遭遇罕见的视觉概念和复杂的时空动态。由于对无限数据分布进行穷举预训练不可行，因此适应新领域的能力至关重要。为弥补这一差距，我们提出AnyGroundBench基准，旨在将STVG评估范式从静态零样本测试转向严格的领域自适应评估。该基准聚焦五个专业领域（动物、工业、体育、手术和公共安全），通过将专家标注的小鼠行为等新采集视频与现有数据集配对，并辅以密集高保真时空标注进行统一整合。尤为关键的是，该基准提供了专用训练子集以系统测量领域适应能力。我们全面评估了15个前沿VLM，在现实计算约束下检验其零样本泛化能力和上下文学习（ICL）能力。最终结果表明，当前模型在应对专业领域时，无论是零样本还是基于ICL的适应均失败，揭示了时空推理中亟需未来研究解决的关键缺陷。

English

Vision-Language Models (VLMs) have demonstrated immense promise in Spatio-Temporal Video Grounding (STVG). However, current evaluation protocols are largely confined to zero-shot assessments on general, daily-life benchmarks. This creates a critical disconnect from real-world applications in specialized fields, where models inevitably encounter rare visual concepts and complex spatio-temporal dynamics. Since exhaustive pre-training across infinite data distributions is infeasible, the ability to adapt to novel domains is essential. To bridge this gap, we introduce AnyGroundBench, a domain-adaptation benchmark designed to shift the STVG evaluation paradigm from static zero-shot testing to rigorous domain adaptation. Targeting five specialized domains (animal, industry, sports, surgery, and public security), AnyGroundBench pairs newly captured videos such as expert-annotated mouse behaviors with established datasets, unifying them through dense, high-fidelity spatio-temporal annotations. Crucially, the benchmark provides dedicated training subsets to systematically measure domain adaptability. We extensively evaluate 15 state-of-the-art VLMs, assessing their zero-shot generalization and In-Context Learning (ICL) capabilities under practical computational constraints. Ultimately, our findings reveal that current models fail in both zero-shot and ICL-based adaptation when confronted with specialized domains, exposing critical flaws in spatio-temporal reasoning that future research must address.