Skin Lesion Classification Using Self-Supervised Learning with DINO

This project explores the use of the DINO framework for skin lesion classification. It covers the development of self-supervised AI algorithms, prior research on skin disease classification, and the acquisition of high-resolution total-body images. The details include the utilization of DINO for feature extraction, advanced data augmentation techniques, and training the model using the HAM10000 dataset. The examples highlight validation with real-world data, visualization of attention maps, ...