Shift-Aware Identity-Guided Latent Refinement for Referring Audio--Visual Segmentation
作者:Kun Li, Sami Sebastian Brandt, Michael Ying Yang · 单位:University of Liverpool, UK, IT University of Copenhagen, DK, University of Bath, UK · 会议/期刊:arXiv preprint · 方向:cs.CV · 发布日期:2026-10-12