Investigating Modality Bias in Audio Visual Video Parsing

TL;DR

This paper analyzes modality bias in audio-visual video parsing models and proposes an improved feature aggregation method that enhances detection accuracy for audio and visual events.

Contribution

It identifies modality bias issues in the existing HAN model and introduces a variant of feature aggregation that improves F-scores in AVVP tasks.

Findings

Modality bias causes certain modalities to be ignored during prediction.

The proposed aggregation method improves F-scores by approximately 1.6-2%.

Enhanced model performance on AVVP benchmarks.

Abstract

We focus on the audio-visual video parsing (AVVP) problem that involves detecting audio and visual event labels with temporal boundaries. The task is especially challenging since it is weakly supervised with only event labels available as a bag of labels for each video. An existing state-of-the-art model for AVVP uses a hybrid attention network (HAN) to generate cross-modal features for both audio and visual modalities, and an attentive pooling module that aggregates predicted audio and visual segment-level event probabilities to yield video-level event probabilities. We provide a detailed analysis of modality bias in the existing HAN architecture, where a modality is completely ignored during prediction. We also propose a variant of feature aggregation in HAN that leads to an absolute gain in F-scores of about 2% and 1.6% for visual and audio-visual events at both segment-level and event-level, in comparison to the existing HAN model.