ChatVTG: Video Temporal Grounding via Chat with Video Dialogue Large Language Models

TL;DR

This work presents ChatVTG, a novel approach that utilizes Video Dialogue Large Language Models (LLMs) for zero-shot video temporal grounding, and surpasses the performance of current zero-shot methods.

Abstract

Video Temporal Grounding (VTG) aims to ground specific segments within an untrimmed video corresponding to the given natural language query. Existing VTG methods largely depend on supervised learning and extensive annotated data, which is labor-intensive and prone to human biases. To address these challenges, we present ChatVTG, a novel approach that utilizes Video Dialogue Large Language Models (LLMs) for zero-shot video temporal grounding. Our ChatVTG leverages Video Dialogue LLMs to generate multi-granularity segment captions and matches these captions with the given query for coarse temporal grounding, circumventing the need for paired annotation data. Furthermore, to obtain more precise temporal grounding results, we employ moment refinement for fine-grained caption proposals. Extensive experiments on three mainstream VTG datasets, including Charades-STA, ActivityNet-Captions, and TACoS, demonstrate the effectiveness of ChatVTG. Our ChatVTG surpasses the performance of current zero-shot methods.

Figures (3)

• Figure 1: Comparison between previous methods based on Video LLMs and our approach: Previous methods require fully supervised training of the Video LLM, whereas our method does not require training and can provide temporal grounding zero-shot.
• Figure 2: Pipeline of ChatVTG. Our method pipeline primarily consists of three components: (a) Instruction-Refined Video Captioning; (b) Query-Caption Matching; and (c) Moment Refinement. Best viewed in color.
• Figure 3: Above the dashed line are successful examples on the Charades-STA dataset, and below are the failed ones.