SIEDOB: Semantic Image Editing by Disentangling Object and Background | Read Paper on Bytez