홈페이지 · 논문 1편
Ref-AVS seeks to segment objects within the visual domain based on expressions containing multimodal cues.