You were saying? -- Spoken Language in the V3C Dataset

TL;DR

This paper analyzes the distribution of spoken language in the V3C dataset using automatic transcripts, highlighting its prevalence and implications for video retrieval tasks like known-item search.

Contribution

It provides the first detailed analysis of spoken language distribution in the V3C dataset, emphasizing its significance for retrieval applications.

Findings

Large portion of dataset contains spoken language

Automatic transcripts enable quick and accurate description

Implications for retrieval tasks such as known-item search

Abstract

This paper presents an analysis of the distribution of spoken language in the V3C video retrieval benchmark dataset based on automatically generated transcripts. It finds that a large portion of the dataset is covered by spoken language. Since language transcripts can be quickly and accurately described, this has implications for retrieval tasks such as known-item search.