Segmentation and Classification of Skin Cancer Using K-means Clustering and EfficientNetB0 Model

Abstract

AbstractNowadays, skin cancer is the widely recognized cancer all over the world. As the spreading rate of skin cancer is increasing day by day, so, there is a need to develop a technique that can detect skin cancer at an early stage. These days, deep learning has attained outstanding success for the detection and diagnosis of cancers. In this paper, a transfer learning-based EfficientNetB0 model is improved by adding one average pooling layer, one dropout layer, one batch normalization and one dense layer with softmax activation function. The proposed model has been simulated using the Kaggle database. The training and calculation are done with different hyper parameters such as batch size, optimizer and epochs. The data augmentation technique is applied to solve the problem of less amount of images. The proposed model has attained 87% accuracy on Adam optimizer with 32 batch size and 30 epochs.KeywordsClassificationK-meansBenignMalignantData augmentationClusteringSkin cancerKaggleEfficientNetB0