Learning What To Hear: Boosting Sound-Source Association For Robust Audiovisual Instance Segmentation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Seo, Jinbae, Kwon, Hyeongjun, Kim, Kwonyoung, Lee, Jiyoung, Sohn, Kwanghoon
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908795125366784
author Seo, Jinbae
Kwon, Hyeongjun
Kim, Kwonyoung
Lee, Jiyoung
Sohn, Kwanghoon
author_facet Seo, Jinbae
Kwon, Hyeongjun
Kim, Kwonyoung
Lee, Jiyoung
Sohn, Kwanghoon
contents Audiovisual instance segmentation (AVIS) requires accurately localizing and tracking sounding objects throughout video sequences. Existing methods suffer from visual bias stemming from two fundamental issues: uniform additive fusion prevents queries from specializing to different sound sources, while visual-only training objectives allow queries to converge to arbitrary salient objects. We propose Audio-Centric Query Generation using cross-attention, enabling each query to selectively attend to distinct sound sources and carry sound-specific priors into visual decoding. Additionally, we introduce Sound-Aware Ordinal Counting (SAOC) loss that explicitly supervises sounding object numbers through ordinal regression with monotonic consistency constraints, preventing visual-only convergence during training. Experiments on AVISeg benchmark demonstrate consistent improvements: +1.64 mAP, +0.6 HOTA, and +2.06 FSLA, validating that query specialization and explicit counting supervision are crucial for accurate audiovisual instance segmentation.
format Preprint
id arxiv_https___arxiv_org_abs_2509_22740
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning What To Hear: Boosting Sound-Source Association For Robust Audiovisual Instance Segmentation
Seo, Jinbae
Kwon, Hyeongjun
Kim, Kwonyoung
Lee, Jiyoung
Sohn, Kwanghoon
Audio and Speech Processing
Artificial Intelligence
Multimedia
Sound
Audiovisual instance segmentation (AVIS) requires accurately localizing and tracking sounding objects throughout video sequences. Existing methods suffer from visual bias stemming from two fundamental issues: uniform additive fusion prevents queries from specializing to different sound sources, while visual-only training objectives allow queries to converge to arbitrary salient objects. We propose Audio-Centric Query Generation using cross-attention, enabling each query to selectively attend to distinct sound sources and carry sound-specific priors into visual decoding. Additionally, we introduce Sound-Aware Ordinal Counting (SAOC) loss that explicitly supervises sounding object numbers through ordinal regression with monotonic consistency constraints, preventing visual-only convergence during training. Experiments on AVISeg benchmark demonstrate consistent improvements: +1.64 mAP, +0.6 HOTA, and +2.06 FSLA, validating that query specialization and explicit counting supervision are crucial for accurate audiovisual instance segmentation.
title Learning What To Hear: Boosting Sound-Source Association For Robust Audiovisual Instance Segmentation
topic Audio and Speech Processing
Artificial Intelligence
Multimedia
Sound
url https://arxiv.org/abs/2509.22740