1 paper
Xingchen Li, Jun Xiao, Guikun Chen +4
Today's scene graph generation (SGG) models typically require abundant manual annotations to learn new predicate types. Therefore, it is difficult to apply them to real-world appli…