Multi-event Video-Text Retrieval

TL;DR

This paper introduces the Multi-event Video-Text Retrieval (MeVTR) task to handle videos with multiple events and proposes a simple yet effective model, Me-Retriever, that outperforms existing models in this new setting.

Contribution

The paper defines the MeVTR task for multi-event videos and proposes the Me-Retriever model with a novel MeVTR loss, addressing a gap in current video-text retrieval methods.

Findings

Me-Retriever outperforms existing models on MeVTR benchmarks.

The proposed model effectively handles videos with multiple events.

The work establishes a new baseline for multi-event video-text retrieval.

Abstract

Video-Text Retrieval (VTR) is a crucial multi-modal task in an era of massive video-text data on the Internet. A plethora of work characterized by using a two-stream Vision-Language model architecture that learns a joint representation of video-text pairs has become a prominent approach for the VTR task. However, these models operate under the assumption of bijective video-text correspondences and neglect a more practical scenario where video content usually encompasses multiple events, while texts like user queries or webpage metadata tend to be specific and correspond to single events. This establishes a gap between the previous training objective and real-world applications, leading to the potential performance degradation of earlier models during inference. In this study, we introduce the Multi-event Video-Text Retrieval (MeVTR) task, addressing scenarios in which each video contains multiple different events, as a niche scenario of the conventional Video-Text Retrieval Task. We present a simple model, Me-Retriever, which incorporates key event video representation and a new MeVTR loss for the MeVTR task. Comprehensive experiments show that this straightforward framework outperforms other models in the Video-to-Text and Text-to-Video tasks, effectively establishing a robust baseline for the MeVTR task. We believe this work serves as a strong foundation for future studies. Code is available at https://github.com/gengyuanmax/MeVTR.