The segmentation outcomes were being then proposed to locate a possible boundary region for each item, which was combined with the initial images for the fine segmentation with the objects. We validated the made method on two general public datasets ( With boundary uncertainty maps and primary en