Machine Learning Based Skin Disease Classification

Abstract

Skin is the most sensitive and the largest organ in our body. Globally more than 35 million people are regularly affected by many types of skin-related diseases. Skin disease is too much danger in Bangladesh for people's lack of awareness. So, to classify those types of skin-related diseases give a proposed system that is related to image processing arising disease analysis is more claim full as they provide promising results in a short time and also used deep learning-based algorithms such as CNN (Convolutional Neural Network) and machine learning based algorithm SVM, Naive Bayes and K-Nearest Neighbor. Those algorithms are connected with image processing for classification and detection. This proposed method often detects whether the given images are impacted by any sort of disease of the skin or not. Here applies a knowledge set which is a combination of sub-continental data then train the data and also after testing those data we applied this technique to various diseases and finally we devise the greatest method that can easily detect those diseases. The first step is image processing with sub-continental data and the second step is using machine learning and deep learning based on the algorithm “Convolutional Neural Network” which can easily help to detect those samples which is given skin disease or not in a very low-cost way. This system can be used by any class of people who don’t afford the high quality and expensive testing method comparing those algorithms, as a result, we probably achieved 83.86% accuracy by using CNN for the detection of various types of skin disease with disease name. CNN gives better performance among them because it can be detected normal skin as well as abnormal skin with great accuracy.