Dynamic Reflections: Probing Video Representations with Text Alignment

TL;DR

This paper explores the alignment between video and text representations, revealing how data richness affects cross-modal alignment and proposing predictive scaling laws, thereby offering a new zero-shot probing method for video understanding.

Contribution

It is the first comprehensive study of video-text alignment, introducing test-time scaling laws and linking alignment quality to downstream task performance.

Findings

Alignment depends on visual and textual data richness.

Proposed scaling laws predict alignment behavior.

Strong alignment correlates with better downstream performance.

Abstract

The alignment of representations from different modalities has recently been shown to provide insights on the structural similarities and downstream capabilities of different encoders across diverse data types. While significant progress has been made in aligning images with text, the temporal nature of video data remains largely unexplored in this context. In this work, we conduct the first comprehensive study of video-text representation alignment, probing the capabilities of modern video and language encoders. Our findings reveal several key insights. First, we demonstrate that cross-modal alignment highly depends on the richness of both visual (static images vs. multi-frame videos) and text (single caption vs. a collection) data provided at test time, especially when using state-of-the-art video encoders. We propose parametric test-time scaling laws that capture this behavior and show remarkable predictive power against empirical observations. Secondly, we investigate the correlation between semantic alignment and performance on both semantic and non-semantic downstream tasks, providing initial evidence that strong alignment against text encoders may be linked to general-purpose video representation and understanding. Finally, we correlate temporal reasoning with cross-modal alignment providing a challenging test-bed for vision and language models. Overall, our work introduces video-text alignment as an informative zero-shot way to probe the representation power of different encoders for spatio-temporal data. Project page can be found at https://video-prh.github.io/