Since this issue is often addressed through fully supervised image segmentation networks, the training process requires significant human effort and time to annotate pixel-level mask labels.