DiscoVLA: Discrepancy Reduction in Vision, Language, and Alignment for Parameter-Efficient Video-Text Retrieval

TL;DR

DiscoVLA introduces a comprehensive approach to improve video-text retrieval by simultaneously addressing vision, language, and alignment discrepancies in adapting CLIP from images to videos, leading to superior performance.

Contribution

The paper proposes a novel method that reduces all three key discrepancies in video-text retrieval, including image-video feature fusion, pseudo caption generation, and alignment distillation.

Findings

Outperforms previous methods on MSRVTT by 1.5% R@1.

Effectively integrates image and video features for better retrieval.

Enhances alignment accuracy through distillation techniques.

Abstract

The parameter-efficient adaptation of the image-text pretraining model CLIP for video-text retrieval is a prominent area of research. While CLIP is focused on image-level vision-language matching, video-text retrieval demands comprehensive understanding at the video level. Three key discrepancies emerge in the transfer from image-level to video-level: vision, language, and alignment. However, existing methods mainly focus on vision while neglecting language and alignment. In this paper, we propose Discrepancy Reduction in Vision, Language, and Alignment (DiscoVLA), which simultaneously mitigates all three discrepancies. Specifically, we introduce Image-Video Features Fusion to integrate image-level and video-level features, effectively tackling both vision and language discrepancies. Additionally, we generate pseudo image captions to learn fine-grained image-level alignment. To mitigate alignment discrepancies, we propose Image-to-Video Alignment Distillation, which leverages image-level alignment knowledge to enhance video-level alignment. Extensive experiments demonstrate the superiority of our DiscoVLA. In particular, on MSRVTT with CLIP (ViT-B/16), DiscoVLA outperforms previous methods by 1.5% in R@1, reaching a final score of 50.5% R@1. The code is available at https://github.com/LunarShen/DsicoVLA.