Processed ISIC2019 Dataset for Skin Disease Diagnosis and Knowledge Distillation Research

This dataset is a curated and processed version of the ISIC2019 skin lesion dataset, specifically prepared for research on lightweight skin disease classification and knowledge distillation. The dataset includes:

A subset of dermoscopic images from ISIC2019, formatted and resized for training and evaluation.

Corresponding metadata tables containing patient information (e.g., age, sex, lesion location).

Pre-processed CSV files that map image names to diagnostic labels.

Split files (train/val/test) for reproducibility.

Supplementary materials for knowledge distillation tasks, such as teacher-student label mappings and class distribution statistics.

This dataset was used in the development of a real-time skin disease diagnosis system optimized for edge deployment. The processed data enables efficient training of lightweight deep learning models and supports reproducible experimentation.

The original ISIC2019 dataset can be found at: https://challenge2019.isic-archive.com/