TY  - RPRT
TI  - Shift-Aware Identity-Guided Latent Refinement for Referring Audio--Visual Segmentation
AU  - Kun Li
AU  - Sami Sebastian Brandt
AU  - Michael Ying Yang
PY  - 2026
UR  - https://arxiv.org/abs/2610.12783
ID  - 2610.12783
ER  -