Zero-Shot Dense Video Captioning by Jointly Optimizing Text and Moment

TL;DR

ZeroTA enables zero-shot dense video captioning by jointly optimizing a language model and a vision-language contrastive model to localize and describe events without any training data.

Contribution

It introduces a novel zero-shot approach for dense video captioning that aligns language and vision models through joint optimization, eliminating the need for annotated training data.

Findings

Outperforms zero-shot baselines on ActivityNet Captions

Surpasses state-of-the-art few-shot methods

Demonstrates robustness in out-of-domain scenarios

Abstract

Dense video captioning, a task of localizing meaningful moments and generating relevant captions for videos, often requires a large, expensive corpus of annotated video segments paired with text. In an effort to minimize the annotation cost, we propose ZeroTA, a novel method for dense video captioning in a zero-shot manner. Our method does not require any videos or annotations for training; instead, it localizes and describes events within each input video at test time by optimizing solely on the input. This is accomplished by introducing a soft moment mask that represents a temporal segment in the video and jointly optimizing it with the prefix parameters of a language model. This joint optimization aligns a frozen language generation model (i.e., GPT-2) with a frozen vision-language contrastive model (i.e., CLIP) by maximizing the matching score between the generated text and a moment within the video. We also introduce a pairwise temporal IoU loss to let a set of soft moment masks capture multiple distinct events within the video. Our method effectively discovers diverse significant events within the video, with the resulting captions appropriately describing these events. The empirical results demonstrate that ZeroTA surpasses zero-shot baselines and even outperforms the state-of-the-art few-shot method on the widely-used benchmark ActivityNet Captions. Moreover, our method shows greater robustness compared to supervised methods when evaluated in out-of-domain scenarios. This research provides insight into the potential of aligning widely-used models, such as language generation models and vision-language models, to unlock a new capability: understanding temporal aspects of videos.