Comparison of KNN and SVM Methods for Melanoma Classification

Melanomas are a type of skin cancer that have a high mortality rate, and early detection is essential to improve patient survival. This article presents a comparison between two machine learning algorithms, KNN and SVM, applied to image processing to classify lesions using images. The methodology of this project was divided into two parts: preprocessing and feature extraction. In the preprocessing, the images were segmented using the Otsu method. For feature extraction, the ABCD test was considered, analyzing the parameters of asymmetry, border, and color. Then, the two machine learning models were trained for comparison. The results showed an accuracy of 58.96% for KNN and 64.09% for SVM.