Skin Disease Classification Using Deep Learning CNN

Abstract

Skin disease is one of the large numbers of spread diseases in the world. Its diagnoses are very difficult because of its difficulties in the skin texture, presence of hair on skin and colour. Thus, it is required to develop an efficient method for the diagnosis of skin disease based on deep learning in order to increase the accuracy of diagnosis for different skin types. Now a days, deep learning techniques are more popular in medical diagnosis system. This work focuses on skin disease prediction using deep learning technique. For the implementation, the dataset used for the skin disease prediction is the ISIC dataset with 9 category of skin diseases. For the classification, the deep learning algorithm used is Visual Geometry Group 19 (VGG 19). VGG 19 is a pretrained neural network, which can be used for the detection of skin disease detection using the concept of transfer learning. For that the neural network extracts image features from the skin images. The extracted features are used for the detection of skin disease within the 7 classes of skin diseases.