一作ECCV 2026PosterFeVOS: Foresight Expression Video Object SegmentationKehan Lan · Kaining Ying · Henghui Ding多模态Video Understanding, Generation and Multimodal LearningOpenReview