Retrieving and Highlighting Action with Spatiotemporal Reference

TL;DR

This paper introduces a novel framework for jointly retrieving and visually highlighting actions in untrimmed videos by leveraging weak supervision from captions, enabling fine-grained spatiotemporal localization.

Contribution

It proposes a new action highlighting task and a framework that generates spatiotemporal relevance maps conditioned on captions, improving retrieval and localization in videos.

Findings

Improves retrieval recall by 2-3% on MSR-VTT dataset.

Generates diverse action-specific spatiotemporal maps.

Enhances understanding of actions beyond traditional saliency methods.

Abstract

In this paper, we present a framework that jointly retrieves and spatiotemporally highlights actions in videos by enhancing current deep cross-modal retrieval methods. Our work takes on the novel task of action highlighting, which visualizes where and when actions occur in an untrimmed video setting. Action highlighting is a fine-grained task, compared to conventional action recognition tasks which focus on classification or window-based localization. Leveraging weak supervision from annotated captions, our framework acquires spatiotemporal relevance maps and generates local embeddings which relate to the nouns and verbs in captions. Through experiments, we show that our model generates various maps conditioned on different actions, in which conventional visual reasoning methods only go as far as to show a single deterministic saliency map. Also, our model improves retrieval recall over our baseline without alignment by 2-3% on the MSR-VTT dataset.