1 paper
Xiangyu Zheng, Wanyun Li, Songcheng He +3
Recent mainstream unsupervised video object segmentation (UVOS) motion-appearance approaches use either the bi-encoder structure to separately encode motion and appearance features…