Skin Cancer Detection using Support Vector Machine with Histogram of Oriented Gradients Features

This research work proposes an efficient skin cancer detection technique based on Support Vector Machine (SVM) with Histogram of Oriented Gradients (HOG) features. In this, skin cancer images from ISIC 2018 (International Skin Imaging Collaboration 2018) dataset are converted into gray scale and pre-processed using the median filter. The image resampling technique is then applied to rebalance the class distribution. The HOG features are extracted from these preprocessed images. After, the Radial Basis Function (RBF) kernel based SVM classification method is used to classify these extracted HOG features for detecting cancer class labels. These predicted class labels are compared with original labels for performing the evaluation. This proposed method is tested using and achieves 76% accuracy, 85% specificity, 84% precision, 76% recall and 75% F1-score.