Themes Informed Audio-visual Correspondence Learning

TL;DR

This paper introduces a novel theme-informed framework for audio-visual correspondence learning tailored for short-term user-generated videos, demonstrating significant improvements on a large new dataset of advertisement videos.

Contribution

The paper proposes new principles and a framework that incorporate video themes into AVC learning, along with releasing a large annotated corpus for evaluation.

Findings

Outperformed baseline by 23.15% absolute difference

Introduced a new large-scale UGV dataset with 85,432 videos

Demonstrated effectiveness of theme-informed AVC approach

Abstract

The applications of short-term user-generated video (UGV), such as Snapchat, and Youtube short-term videos, booms recently, raising lots of multimodal machine learning tasks. Among them, learning the correspondence between audio and visual information from videos is a challenging one. Most previous work of the audio-visual correspondence(AVC) learning only investigated constrained videos or simple settings, which may not fit the application of UGV. In this paper, we proposed new principles for AVC and introduced a new framework to set sight of videos' themes to facilitate AVC learning. We also released the KWAI-AD-AudVis corpus which contained 85432 short advertisement videos (around 913 hours) made by users. We evaluated our proposed approach on this corpus, and it was able to outperform the baseline by 23.15% absolute difference.