Design And Implementation of a Real Time Skin Cancer Detection System

Abstract

Skin Cancer Detection System Application (SKDSA): is an application that depends on a transfer learning technique that includes a deep learning model (resnet50) and machine learning model (SVM) for the classification of skin images, this technique is used for the early detection method of skin cancer, due to rise cases of skin cancer around the world at a rate more than 1.5 million new cases estimated in 2020 the need increased for a new method for detection of skin cancer with low cost and time.

In this system, the user must provide the image of the affected area, then the input image undergoes pre-processing which includes filtering to remove noise, segmentation to extract the lesion, and then feature extraction by resnet50 network to extract image features, and finally classifier (SVM) to detect the affected area and give predicted classification. 

Our project aims to classifying skin images if the image contains skin cancer diseases or not. Enables users to detect and identify types of skin cancer diseases located in images with high accuracy of classification and low cost and less time than other methods of classification such as biopsy. It is an effective method to use for the early detection of real time skin images or loaded skin image on computer.