Breaking the "Object" in Video Object Segmentation | Read Paper on Bytez