Skin Cancer Classification Model Based on VGG 19 and Transfer Learning
Nour Abuared, Alavikunhu Panthakkan, Mina Al-Saad, Saad Amin, Wathiq Mansoor
Abstract
Skin cancer is a concerning health issue with yearly increasing numbers. Detecting and classifying cancer type is problematic, especially since patients have to undergo several diagnosis over lengthy periods of time, which hinders early treatment and survival chances. With the aid of digital image processing, features can be extracted to identify skin cancer and its different types. Convolutional Neural Networks (CNNs) recently emerged as powerful autonomous feature extractors, and they have high potential to achieve high accuracy with skin cancer diagnosis. In this paper, two cancer types in addition to one non-cancer type taken from Human Against Machine (HAM10000) dataset are classified using CNN model based on VGG 19 and Transfer Learning technique. The training strategy is explained, tested, and evaluated by calculating the network's overall accuracy and loss.