Virtual Multi-Modality Self-Supervised Foreground Matting for
Human-Object Interaction
Bo Xu1 , Han Huang1 , Cheng Lu2 , Ziwen Li1,3 and Yandong Guo1,* GT
1
OPPO Research Institute, 2 Xmotors, 3 University of California, San Diego
yandong.guo@live.com
Abstract
Most existing human matting algorithms tried to sepa-
rate pure human-only foreground from the background. In
this paper, we propose a Vir ...


雷达卡




京公网安备 11010802022788号







