Minimal Clips, Maximum Salience: Long Video Summarization via Key Moment Extraction

TL;DR

This paper introduces a lightweight clip selection method that extracts key moments from long videos to generate effective multimodal summaries, maintaining low computational costs and high relevance.

Contribution

It presents a novel approach combining short clip generation and large language models for efficient, relevant video summarization of lengthy content.

Findings

Achieves near-reference summarization performance using less than 6% of the video.

Captures more relevant video information than random clip selection.

Maintains low computational cost with lightweight models.

Abstract

Vision-Language Models (VLMs) are able to process increasingly longer videos. Yet, important visual information is easily lost throughout the entire context and missed by VLMs. Also, it is important to design tools that enable cost-effective analysis of lengthy video content. In this paper, we propose a clip selection method that targets key video moments to be included in a multimodal summary. We divide the video into short clips and generate compact visual descriptions of each using a lightweight video captioning model. These are then passed to a large language model (LLM), which selects the K clips containing the most relevant visual information for a multimodal summary. We evaluate our approach on reference clips for the task, automatically derived from full human-annotated screenplays and summaries in the MovieSum dataset. We further show that these reference clips (less than 6% of the movie) are sufficient to build a complete multimodal summary of the movies in MovieSum. Using our clip selection method, we achieve a summarization performance close to that of these reference clips while capturing substantially more relevant video information than random clip selection. Importantly, we maintain low computational cost by relying on a lightweight captioning model.