Open AccessDerma Net: An Automated Skin Lesion Analyzer using CNN with Adaptive Learning
Author(s) -
Katakam Koushik,
Arjun Krishna,
D Sai Tharun
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.f1107.0486s419
Subject(s) - computer science , dropout (neural networks) , artificial neural network , artificial intelligence , regularization (linguistics) , deep learning , machine learning , computation , algorithm
In this paper we are going to develop an automated skin lesion analyzer that can take affected skin lesion image from user and predict or approximate 3 skin diseases with 95% accuracy. To accomplish this goal we are going to use Neural Networks as they are the best data driven models with top most accuracy in all the fields they have been experimented till now. Since Neural Network models also need huge computation power to train the model on the input data and also to predict the output we are going to use a computationally less intensive architecture that can work even on hand held mobiles and embedded systems. To further featuring our model we have added dropout techniques for model regularization and adaptive learning rates to achieve global minima with ease even with the presence of plateaus. At last we will deploy a production level web application to serve users across the world