Few-shot learning for rare skin disease classification via adaptive distribution calibration.

The classification of rare skin diseases faces significant data scarcity challenges due to the difficulty in acquiring clinical samples and the high cost of annotation, which severely hinders the training of deep neural network-based models. Few-shot…

Preview only. Read the full abstract at the source