Fitz Selected: Seven-Class Skin Lesion Subset for Domain Adaptation Research

This dataset, titled Fitz Selected, is a curated subset of the Fitzpatrick17k dataset containing seven classes of skin lesion images aligned with the ISIC 2019 taxonomy. It was prepared for research on unsupervised domain adaptation (UDA) in multi-class skin disease classification. Each image is organized into one of the following classes: actinic keratosis (ak), basal cell carcinoma (bcc), benign keratosis (bkl), dermatofibroma (df), melanoma in situ (mei), nevus (nv), and squamous cell carcinoma (scc). This dataset was created to support reproducible experiments in skin lesion classification and domain adaptation studies.