An automated detection of Scabies skin disease Using Image Processing and CNN

Abstract

In every year, human faces several types of skin diseases. Bacteria, fungal infections, viruses, allergies, etc. are the reason for different types of skin diseases. Many people don’t show interest in identifying or treatment for skin diseases because of the cost of diagnosis which is expensive and lack of concentration. Image processing is one the easiest way to detect skin diseases by classifying the image of the affected area. A dermatologist can easily identify the problem by using the image processing technique. Feature extraction helps to classify skin diseases effectively. In this paper, we focused on Scabies which is one of the most common skin diseases. We proposed a methodology using image processing and CNN to detect scabies. Here thresholding is used for segmenting the affected area and CNN is used for classifying images. We used different types of data augmentation techniques for increasing data where each of the images was considered unique. Our proposed method is simple and easy to use. In this research, RGB images are used and we made a dataset where all images are collected from different sources in the internet for training the system. The proposed methodology can detect scabies with an accuracy of 97.25%.