PolySmart @ TRECVid 2024 Video Captioning (VTT)

TL;DR

This paper explores fine-tuning Vision-Language Models like LLaVA for video captioning at TRECVid 2024, showing significant improvements in description quality and relevance for video content.

Contribution

It demonstrates the effectiveness of domain-specific fine-tuning of VLMs for the Video-To-Text task, advancing the state-of-the-art in video captioning.

Findings

Fine-tuning improves description accuracy and relevance.

Fine-tuned models outperform baseline VLMs.

Enhanced linguistic consistency in generated captions.

Abstract

In this paper, we present our methods and results for the Video-To-Text (VTT) task at TRECVid 2024, exploring the capabilities of Vision-Language Models (VLMs) like LLaVA and LLaVA-NeXT-Video in generating natural language descriptions for video content. We investigate the impact of fine-tuning VLMs on VTT datasets to enhance description accuracy, contextual relevance, and linguistic consistency. Our analysis reveals that fine-tuning substantially improves the model's ability to produce more detailed and domain-aligned text, bridging the gap between generic VLM tasks and the specialized needs of VTT. Experimental results demonstrate that our fine-tuned model outperforms baseline VLMs across various evaluation metrics, underscoring the importance of domain-specific tuning for complex VTT tasks.