VideoLoom: A Video Large Language Model for Joint Spatial-Temporal Understanding

TL;DR

VideoLoom introduces a unified large language model for comprehensive spatial-temporal video understanding, leveraging a new dataset and benchmark to achieve state-of-the-art results across multiple video understanding tasks.

Contribution

The paper presents VideoLoom, a novel Video LLM with a curated dataset LoomData-8.7k and a comprehensive benchmark LoomBench for joint spatial-temporal video understanding.

Findings

Achieves 63.1 J&F on ReVOS for video object segmentation

Attains 48.3 R1@0.7 on Charades-STA for temporal grounding

Sets new standards in multimodal video understanding

Abstract

This paper presents VideoLoom, a unified Video Large Language Model (Video LLM) for joint spatial-temporal understanding. To facilitate the development of fine-grained spatial and temporal localization capabilities, we curate LoomData-8.7k, a human-centric video dataset with temporally grounded and spatially localized captions. With this, VideoLoom achieves state-of-the-art or highly competitive performance across a variety of spatial and temporal benchmarks (e.g., 63.1 J&F on ReVOS for referring video object segmentation, and 48.3 R1@0.7 on Charades-STA for temporal grounding). In addition, we introduce LoomBench, a novel benchmark consisting of temporal, spatial, and compositional video-question pairs, enabling a comprehensive evaluation of Video LLMs from diverse aspects. Collectively, these contributions offer a universal and effective suite for joint spatial-temporal video understanding, setting a new standard in multimodal intelligence.