XSSR: Cross-Domain Self-Supervised Selection for Medical Image Annotation

Arxiv CS.CV (computer vision)

XSSR introduces a cross-domain self-supervised representative selection method to reduce annotation costs for medical image segmentation. It selects the most informative samples from the target domain without requiring labels, addressing domain shifts due to differences in imaging equipment or clinical sites.

computer-visionself-supervised-learning

Arxiv CS.CV (computer vision)·Byunghyun KoAleksei AnisimovKobe KeSuhas BharthepudeJeongkyu Lee

7 days ago