Wolf: Dense Video Captioning with a World Summarization Framework

TL;DR

Wolf is a novel framework for dense video captioning that combines vision-language models and a new evaluation metric, CapScore, to improve caption quality across diverse domains and surpass existing methods.

Contribution

The paper introduces Wolf, a mixture-of-experts framework utilizing multiple vision-language models and a new LLM-based metric, CapScore, for enhanced and comprehensive video captioning.

Findings

Wolf outperforms state-of-the-art methods and commercial solutions in caption quality.

CapScore correlates well with human judgment of caption quality.

The framework is validated across datasets in autonomous driving, scenes, and robotics.

Abstract

We propose Wolf, a WOrLd summarization Framework for accurate video captioning. Wolf is an automated captioning framework that adopts a mixture-of-experts approach, leveraging complementary strengths of Vision Language Models (VLMs). By utilizing both image and video models, our framework captures different levels of information and summarizes them efficiently. Our approach can be applied to enhance video understanding, auto-labeling, and captioning. To evaluate caption quality, we introduce CapScore, an LLM-based metric to assess the similarity and quality of generated captions compared to the ground truth captions. We further build four human-annotated datasets in three domains: autonomous driving, general scenes, and robotics, to facilitate comprehensive comparisons. We show that Wolf achieves superior captioning performance compared to state-of-the-art approaches from the research community (VILA1.5, CogAgent) and commercial solutions (Gemini-Pro-1.5, GPT-4V). For instance, in comparison with GPT-4V, Wolf improves CapScore both quality-wise by 55.6% and similarity-wise by 77.4% on challenging driving videos. Finally, we establish a benchmark for video captioning and introduce a leaderboard, aiming to accelerate advancements in video understanding, captioning, and data alignment. Webpage: https://wolfv0.github.io/.