Performance and Accuracy Enhancement During Skin Disease Detection in Deep Learning

Abstract

Epidermolysis bullosa is a type of skin cancer that is consistently ranked as among the worst diseases in the world. Accurate categorization of skin lesions in their early stages may assist during clinical deliberation, hence increasing the possibility of a cure before cancer starts. The components that are often affected by direct sunlight are the ones most likely to acquire a form of skin cancer. These include the head, and body parts that are more obvious in men than women are the legs and feet. However, it can also develop on parts of your body that are rarely subjected to air and light, such as your hands and feet. Researchers today are thinking about using deep learning to identify skin cancer more quickly and accurately. Compression operations have been performed to boost performance, while a hybrid deep learning model has enhanced accuracy metrics like recall, precision, and f1-score.