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

Key points

• 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).1
• 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.1
• To facilitate research on Event-STVG, we construct a large-scale benchmark comprising 4,016 annotated event occurrences across soccer and seven additional domains.1
• 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.1

Sources (1)

• [1]Event-STVG: Spatio-Temporal Video Grounding for Long-Form Event Reasoning
  Hugging Face Daily Papers · Oct 15, 12:00 AM

  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).

Extractive summary: sentences quoted from the sources.