Event-STVG: Spatio-Temporal Video Grounding for Long-Form Event Reasoning

Computer Science > Computer Vision and Pattern Recognition arXiv:2610.22444 (cs) [Submitted on 15 Oct 2026] Title: Event-STVG: Spatio-Temporal Video Grounding for Long-Form Event Reasoning Authors: Yizhe Zhang, Xujia Wang, Songqi Tang, Xujie Tang, Leo Jiahao Jiang, Jiacheng Wang, Bowen Qu, JinBo Li, Qin Wang, Ji Qi, Lei Hou, Bin Xu, Juanzi Li.

计算机科学 > 计算机视觉与模式识别 arXiv:2610.22444 (cs) [提交于 2026 年 10 月 15 日] 标题：Event-STVG：用于长时序事件推理的时空视频定位。作者：Yizhe Zhang, Xujia Wang, Songqi Tang, Xujie Tang, Leo Jiahao Jiang, Jiacheng Wang, Bowen Qu, JinBo Li, Qin Wang, Ji Qi, Lei Hou, Bin Xu, Juanzi Li。

Abstract: Spatio-Temporal Video Grounding (STVG) aims to jointly localize a target in space and time given a natural-language query. Despite continued advances, existing STVG benchmarks remain largely centered on localizing targets through instance-specific descriptions, leaving them largely decoupled from long-form reasoning increasingly emphasized in multimodal large language models (MLLMs).

摘要：时空视频定位（STVG）旨在根据自然语言查询，在空间和时间上联合定位目标。尽管进展不断，但现有的 STVG 基准测试主要集中于通过实例特定描述来定位目标，这使得它们与多模态大语言模型（MLLMs）中日益强调的长时序推理在很大程度上脱节。

In this paper, we introduce Event-STVG, a task that shifts STVG from locating what is described toward event-level reasoning by unifying long-form context, repeated-event reasoning, event localization, role inference, and multi-target grounding. To facilitate research on Event-STVG, we construct a large-scale benchmark comprising 4,016 annotated event occurrences across soccer and seven additional domains.

在本文中，我们引入了 Event-STVG，这是一项将 STVG 从定位描述对象转向事件级推理的任务，通过整合长时序上下文、重复事件推理、事件定位、角色推断和多目标定位来实现。为了促进 Event-STVG 的研究，我们构建了一个大规模基准测试，包含足球及其他七个领域中 4,016 个已标注的事件实例。

Experiments with state-of-the-art MLLMs reveal substantial difficulty, with the strongest model achieving only 5.4% m-vIoU at the 10-minute horizon. Across domains, we consistently observe confusion among repeated same-type events, spatial grounding bottlenecks, and rare success in jointly grounding all required roles.

对最先进的 MLLMs 进行的实验揭示了巨大的难度，最强的模型在 10 分钟的时间跨度内仅达到 5.4% 的 m-vIoU。在各个领域中，我们持续观察到重复的同类事件之间的混淆、空间定位瓶颈，以及在联合定位所有所需角色方面鲜有成功。

Event-STVG establishes a new challenge for visual understanding in MLLMs, calling for future models that can jointly reason about complex events across space and time. Code and annotations are available at this https URL.

Event-STVG 为 MLLMs 中的视觉理解设立了新的挑战，呼吁未来的模型能够跨越空间和时间对复杂事件进行联合推理。代码和标注可在该网址获取。