Skin disease detection and classification using deep learning

Abstract

Skin Diseases have been the primary focus of this study, as they are one of the most lethal diseases if not diagnosed and treated early. The research will enable the fields of Medical Science and Computer Science to collaborate in order to save lives. Although Machine Learning, Deep Learning, and Image Processing have been utilized previously to treat skin diseases, we are attempting to improve the accuracy of this work by implementing new models of Image Processing and Deep Learning. The purpose of this research is to demonstrate how to accurately diagnose Skin diseases at an early stage using the optimum model. Here we have used three distinct neural models to classify a custom dataset. We further analyzed the accuracy of the MobileNetV2, InceptionV3, and ResNetV2 to come up with an optimized model that can be configured further to a mobile application for vast use. We built the architecture on more than 1450 images representing nine distinct skin disorders in comparison with fresh skin. We carefully compared our data and classified it based on the images of our customized dataset. Finally, we determined the nine diseases with a 96.77% accuracy with the help of MobileNetV2 which is our ideal model for the goal we want to achieve.