Classification of Skin Disease from Skin images using Transfer Learning Technique
Honey Janoria, Jasmine Minj, Pooja Patre
Abstract
Skin Diseases are one of the most common ailments in today's era. Due to skin disease some small circular or random shaped area can be seen on the patient's skin. This disease may be very dangerous in some situations when it converts to skin cancer. Here in this article, some deep learning-based approaches were discussed which can be used to extract features from the different skin cancer images, and then these features are used to detect the type of skin disease using some machine learning classifiers. For our experiments, a transfer learning model is developed in which, for feature extraction the VGG-16 layer CNN architecture can extract 1000 features from the input image and, for the classification purpose, and used a support vector machine, decision tree, linear discriminate analysis, and K-Nearest Neighbor algorithm as they are best suited for linear classification. The experiments have been performed on well known public datasets of ISIC. The experimental results show that the highest accuracy of 99% was achieved by using the VGG 16 CNN model with the K-Nearest Neighbor algorithm.