3rd Place of MeViS-Audio Track of the 5th PVUW: VIRST-Audio

TL;DR

VIRST-Audio is a framework for audio-based video object segmentation that converts audio to text and uses text-based supervision, with a gating mechanism to improve robustness, achieving competitive results in a challenge.

Contribution

It introduces a novel approach that leverages text-based reasoning for audio-driven segmentation without audio-specific training, enhancing robustness with an existence-aware gating mechanism.

Findings

Achieved 3rd place in PVUW challenge

Demonstrated effective transfer from text to audio scenarios

Improved segmentation stability with gating mechanism

Abstract

Audio-based Referring Video Object Segmentation (ARVOS) requires grounding audio queries into pixel-level object masks over time, posing challenges in bridging acoustic signals with spatio-temporal visual representations. In this report, we present VIRST-Audio, a practical framework built upon a pretrained RVOS model integrated with a vision-language architecture. Instead of relying on audio-specific training, we convert input audio into text using an ASR module and perform segmentation using text-based supervision, enabling effective transfer from text-based reasoning to audio-driven scenarios. To improve robustness, we further incorporate an existence-aware gating mechanism that estimates whether the referred target object is present in the video and suppresses predictions when it is absent, reducing hallucinated masks and stabilizing segmentation behavior. We evaluate our approach on the MeViS-Audio track of the 5th PVUW Challenge, where VIRST-Audio achieves 3rd place, demonstrating strong generalization and reliable performance in audio-based referring video segmentation.