PolySmart @ TRECVid 2024 Video Captioning (VTT)
This work addresses the challenge of producing domain-aligned video captions for applications like video indexing and accessibility, representing an incremental improvement through domain-specific tuning.
The paper tackled the problem of generating natural language descriptions for video content by fine-tuning Vision-Language Models (VLMs) like LLaVA and LLaVA-NeXT-Video on Video-To-Text (VTT) datasets, resulting in improved accuracy, contextual relevance, and linguistic consistency, with the fine-tuned model outperforming baseline VLMs across various evaluation metrics.
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.