TEMPURA: Temporal Event Masked Prediction and Understanding for Reasoning in Action

TL;DR

TEMPURA introduces a novel two-stage training framework that enhances temporal understanding and causal reasoning in videos by reconstructing missing events and segmenting videos into detailed, timestamp-aligned events, outperforming existing models.

Contribution

The paper presents TEMPURA, a new approach combining masked event prediction and video segmentation for improved causal and temporal reasoning in videos, trained on a large-scale dataset.

Findings

Outperforms baseline models on temporal grounding tasks.

Effectively reconstructs missing events and generates causal explanations.

Enhances fine-grained temporal segmentation and understanding.

Abstract

Understanding causal event relationships and achieving fine-grained temporal grounding in videos remain challenging for vision-language models. Existing methods either compress video tokens to reduce temporal resolution, or treat videos as unsegmented streams, which obscures fine-grained event boundaries and limits the modeling of causal dependencies. We propose TEMPURA (Temporal Event Masked Prediction and Understanding for Reasoning in Action), a two-stage training framework that enhances video temporal understanding. TEMPURA first applies masked event prediction reasoning to reconstruct missing events and generate step-by-step causal explanations from dense event annotations, drawing inspiration from effective infilling techniques. TEMPURA then learns to perform video segmentation and dense captioning to decompose videos into non-overlapping events with detailed, timestamp-aligned descriptions. We train TEMPURA on VER, a large-scale dataset curated by us that comprises 1M training instances and 500K videos with temporally aligned event descriptions and structured reasoning steps. Experiments on temporal grounding and highlight detection benchmarks demonstrate that TEMPURA outperforms strong baseline models, confirming that integrating causal reasoning with fine-grained temporal segmentation leads to improved video understanding.