Test of Time: Instilling Video-Language Models with a Sense of Time

TL;DR

This paper investigates the temporal understanding of video-language models, identifies their limitations in grasping simple time relations, and proposes a lightweight adaptation method to improve their temporal awareness without retraining from scratch.

Contribution

It demonstrates the difficulty of existing models in understanding basic temporal relations and introduces a post-pretraining adaptation approach to enhance their temporal sense efficiently.

Findings

Existing models struggle with simple temporal relations.

Post-pretraining improves temporal understanding in downstream tasks.

Encouraging performance gains in tasks requiring high temporal awareness.

Abstract

Modelling and understanding time remains a challenge in contemporary video understanding models. With language emerging as a key driver towards powerful generalization, it is imperative for foundational video-language models to have a sense of time. In this paper, we consider a specific aspect of temporal understanding: consistency of time order as elicited by before/after relations. We establish that seven existing video-language models struggle to understand even such simple temporal relations. We then question whether it is feasible to equip these foundational models with temporal awareness without re-training them from scratch. Towards this, we propose a temporal adaptation recipe on top of one such model, VideoCLIP, based on post-pretraining on a small amount of video-text data. We conduct a zero-shot evaluation of the adapted models on six datasets for three downstream tasks which require varying degrees of time awareness. We observe encouraging performance gains especially when the task needs higher time awareness. Our work serves as a first step towards probing and instilling a sense of time in existing video-language models without the need for data and compute-intense training from scratch.